Journal of Network and Computer Applications 207 (2022) 103476

Contents lists available at ScienceDirect

Journal of Network and Computer Applications

journal homepage: www.elsevier.com/locate/jnca

RPL routing protocol over IoT: A comprehensive survey, recent advances,

insights, bibliometric analysis, recommendations, and future directions
Khalid A. Darabkh a, *, Muna Al-Akhras a, Jumana N. Zomot b, Mohammed Atiquzzaman c
a
Department of Computer Engineering, The University of Jordan, Amman, 11942, Jordan
b
Information Technology Directorate, Ministry of Transport, Amman, 11180, Jordan
c
Presidential Professor of Computer Science, University of Oklahoma Norman, OK, 73019, USA

A R T I C L E I N F O A B S T R A C T

Keywords: In the course of time, the Internet of Things (IoT) has attracted significant research interest. However, IoT devices
Internet of things (IoT) have limited resources available in terms of battery power, processing capacity, memory, bandwidth, etc. In an
Routing protocols attempt to provide connectivity and Internet Protocol version 6 (IPv6) support to IoT devices, the IPv6 routing
RPL
protocol for Low-Power and Lossy Networks (RPL) was officially launched as the standard routing protocol for
RPL challenges
enhancements
IoT in 2012. Despite being reputed and used in diverse applications, several recent studies have revealed RPL’s
and recommendations drawbacks and limitations. The main objective of this work is to help the IoT research community understand all
RPL future directions aspects of RPL. The paper also provides a detailed description of the operation of the RPL protocol. What is more,
this work includes novel and thorough examples, thereby gaining practical knowledge of the pros and cons of this
protocol. In addition, this paper reviews and summarizes relevant RPL-based protocols and conducts compre­
hensive comparisons among them from the perspectives of reliability, robustness, energy efficiency, and flexi­
bility. Technically speaking, after studying and reviewing the majority of the proposed RPL solutions, we are
ultimately capable, in this work, of highlighting all the challenges faced by IoT researchers while enhancing RPL
and providing what is expected to be dealt with professionally. The present work also gives more details about
RPL simulation platforms and RPL applications. Not only to this extent, but rather the historical bibliometric
analysis of RPL, which shows the trends in the area of research to be focused on, has been professionally analyzed
based on RPL challenges over the years 2010 through 2021. To this end, the conclusions and recommendations of
this study are presented along with the effective directions for future RPL, and their applicability. As a result, the
authors believe that this work will be a valuable reference for all RPL researchers and designers.

1. Introduction devices connected with the IoT by the year 2025 (Harbi et al., 2021)
(Darabkh et al., 2020a). Furthermore, it is also projected that in 2025,
Over the past few years, the Internet of Things (IoT) has been the title IoT devices will play an essential role in people’s daily lives (Council,
of the new era of Internet networks (Harb et al., 2017) (Chen et al., 2008) (Darabkh et al., 2019b). The advent of IoT technology has given
2022). The IoT consists of many physical objects that are connected to rise to a new special type of network called Low-Power and Lossy Net­
the Internet, i.e., sensors, vehicles, and devices equipped with micro­ works (LLNs) (Janicijevic et al., 2011). LLNs have a slew of restrictions,
processors, to name a few, that collect and exchange data (Iova et al., including not only resource-constrained devices, which are character­
2016) (Banh et al., 2016). A wide range of applications can benefit from ized by limited memory, power, and computational capabilities, but also
these devices, such as smart cities, healthcare, and the military, which is communication links (i.e., low bandwidth, short transmission range, and
in the interest of mankind (Darabkh et al., 2019a). For that reason, the dynamically changing network topologies due to the mobile node),
IoT is expected to grow to an enormous size over the next few years which cause high packet loss, low end-to-end delay, high energy con­
(Darabkh et al., 2022a) (Darabkh and Amro, 2020). In (Evans, 2011), sumption, and low throughput (Pavkovic et al., 2014) (Darabkh and
Cisco has anticipated that there will be approximately 100 billion Al-Jdayeh, 2019).

* Corresponding author.
E-mail addresses: [email protected] (K.A. Darabkh), [email protected] (M. Al-Akhras), [email protected] (J.N. Zomot), [email protected]
(M. Atiquzzaman).

https://doi.org/10.1016/j.jnca.2022.103476
Received 21 July 2021; Received in revised form 23 June 2022; Accepted 22 July 2022
Available online 17 August 2022
1084-8045/© 2022 Elsevier Ltd. All rights reserved.
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Pertinently, Wireless Sensor Networks (WSNs) play a vital role in the Protocol (CoAP) (Bormann et al., 2012). It enables limited power, tiny,
origination and development of the IoT (Qiu et al., 2019) (Alhasanat and embedded devices for intercommunicating over the Internet
et al., 2020). As a result, it is possible to connect low-end devices with (Ancillotti et al., 2014a).
restricted resources to the Internet and possibly provide innovative Since the standardization of RPL, it has received significant traction
services (Chen et al., 2021) (Darabkh et al., 2017) (Darabkh and Kassab, in both industry and academia (Dhumane and Prasad, 2015). The rea­
2020). Although the IoT is inherited from WSNs, the conventional sons behind this are that it provides efficient and stable routes among
routing protocols used in WSNs became improper for LLNs (Granjal nodes in LLNs, supports Quality of Service (QoS), and is flexible in
et al., 2015) (Darabkh et al., 2019c). Based on that, one of the salient adapting to varied network topologies (Triantafyllou et al., 2018)
challenges in LLNs is to use an efficient routing protocol that meets the (Talwar, 2015) (Seyfollahi et al., 2022) (Mishra et al., 2022). In addi­
requirements of the applications, such as considering low-power IoT tion, RPL employs low-energy mechanisms to support self-organization
devices and short transmission range (Salman and Jain, 2017) (Darabkh and self-healing to control frequent node failures (Ghaleb et al., 2019).
et al., 2021a). In response to this challenge, standardization groups, such Technically speaking, adapting IPv6 has several potential benefits,
as the Internet Engineering Task Force (IETF) and the Institute of Elec­ including universality, scalability, and reliability (Thomson et al.,
trical and Electronics Engineers (IEEE), have defined the Internet Pro­ 2007). Its tremendous addressing range is also suitable for addressing a
tocol version 6 (IPv6) protocol stack. Fig. 1 shows the set of standardized substantial percentage of IoT devices. It is worth mentioning that the
protocols and the corresponding layers (Sebastian and Sivagurunathan, RPL was created with resource-constrained devices in mind, allowing it
2018) (Kassab and Darabkh, 2020). to handle a variety of IoT-LLN applications. Indeed, RPL is a proactive
IEEE 802.15.4 is a well-known and widely used standard that was distance-vector loop-free routing protocol (Sharma and Shukla, 2014).
launched in 2003 and defines how Low-Rate Wireless Personal Area More precisely, RPL creates a Destination Oriented Directed Acyclic
Networks (LR-WPANs) operate and the specifications of the Medium Graph (DODAG) based on a set of node and link metrics and constraints,
Access Control (MAC) and Physical (PHY) layers for LR-WPANs which contribute to designing an Objective Function (OF) that is con­
(Gutierrez et al., 2001) (Darabkh et al., 2020b) (Khalifeh et al., 2019). verted into a single value called the rank, which is mainly responsible for
Moreover, IEEE 802.15.4 can be used for a large diversity of higher-layer selecting a preferred parent of each node. The process continued until
standards, such as Zigbee, Wireless HART, MiWi, ISA100.11a, radio establishing the network by choosing the most efficient routes to the
frequency for consumer electronics, and IPv6 over Low-power Wireless destination (Darabkh and Al-Akhras, 2021).
Personal Area Networks (6LoWPAN). Intending to allow low-end de­ Various research studies have shown that RPL has strict limitations
vices with limited power to connect to the Internet, the IETF created and drawbacks that affect the network performance. For this reason,
6LoWPAN in 2004 (Moritz and Lerche, 2013). Actually, the goal of the several new solutions studied in the context of this paper have recently
6LoWPAN was to find ways to enable effective IPv6 transmission over emerged in an attempt to mitigate existing routing problems. This paper
IEEE 802.15.4 networks. Accordingly, 6LoWPAN is proposed to include is a pioneer in reviewing, systematically, RPL and RPL-based solutions,
an adaptation layer between the (IPv6) network and the (IEEE 802.15.4) besides the visions and recommendations concerning these imple­
data link layer (Aijaz and Aghvami, 2015). To be more specific, this mentations. Thus, the target of this survey lies in satisfactorily pre­
layer has encapsulation and header compression techniques to send and senting RPL by illustrating its operations and demonstrating informative
receive IPv6 packets on IEEE 802.15.4 communication channels (Pal­ examples for building the DODAG with different metrics. Plus, this work
attella et al., 2013). highlights the most current challenges of RPL and the enhanced solu­
The introduction of 6LoWPAN underlined the necessity of further tions, along with the recommendations and future directions. In view of
IPv6-based LLNs routing solutions, and afterwards, in 2008, the IETF that, our work is regarded as a good reference for RPL because it facil­
created a new Working Group (WG) called Routing Over Low-power and itates the process of researching and suggesting innovative RPL
Lossy networks (ROLL). ROLL aims to design these routing solutions that optimizations.
satisfy the routing requirements of LLNs (Gaddour and Koubâa, 2012). It is pertinent to mention that the preliminary findings of this study
In 2012, the ROLL identified the IPv6 routing protocol for LLNs (RPL) as were published in a very simplified version in (Darabkh and Al-Akhras,
a network layer and standard routing protocol designed for LLNs and IoT 2021).
networks (Winter et al., 2012). Likewise, the stack includes the standard
User Datagram Protocol (UDP) in the transport layer. Finally, for the 1.1. Related works
application layer, the employed protocol is the Constrained Application
In the literature, many attractive surveys have provided a concise but
not extensive review of the RPL protocol. Nevertheless, most of these
studies have focused on reviewing the functionality of the RPL and
analyzing the restrictions of the protocol itself rather than evaluating the
various research efforts undertaken to overcome these restrictions.
Moreover, these surveys partially covered various aspects of RPL or only
highlighted a few aspects of this routing protocol. Our aim in this survey
is to take these insights into consideration, and consequently, go beyond
the existing literature. This subsection provides a literature review of
some of these works with a short discussion on the topics covered in
their surveys. It is worth noting that, the related works in this subsection
are compared to our work in Table 1.
In (Gaddour and Koubâa, 2012), the authors reviewed relevant
research efforts related to the implementation, experimentation,
dissemination, optimization, and performance evaluation of the RPL.
They also performed an experimental performance evaluation of the RPL
for different network settings to comprehend the effect of RPL features
on network performance. Although their survey provided a compre­
hensive study of the RPL routing protocol, it did not discuss research
efforts to overcome the restrictions of RPL or did not provide informa­
Fig. 1. IPv6 protocol stack standardized by IETF and IEEE. tive examples of RPL operations. Ben Saad et al. (Ben Saad et al., 2011)

2
 K.A. Darabkh et al.
Table 1
Comparison of our work with other pertinent surveys reviewed in this study in terms of RPL topics covered.
Article Article Title Year Extensive RPL Trickle Routing DODAG Exclusive DODAG Routing Recent RPL Recent Recent Recent Historical Research
specifications algorithm metric/ construction examples local table for applications RPL RPL RPL bibliometric recommendations
constraints comprehensive with repair nodes simulators challenges enhanced analysis
OFs examples different example and root protocols
OFs

Ben Saad et al. Simulation of the RPL 2011 × × × × × × × × ✓ × × × ×

(Ben Saad Routing Protocol for IPv6
et al., 2011) Sensor Networks: two
cases studies
Tsvetko RPL: IPv6 Routing 2011 ★ × ☆ × × × × × × × × × ×
Tsvetkov ( Protocol for Low Power
Tsvetkov, and Lossy Networks
2011)
Gaddour and RPL in a nutshell: A 2012 ✓ × × × × × × × × × × × ×
Koubâa ( survey
Gaddour and
Koubâa,
2012)
Al-Fuqaha Internet of Things: A 2015 ☆ × × × × × × ☆ ☆ ☆ × × ☆
et al. ( Survey on Enabling
Al-Fuqaha Technologies, Protocols
et al., 2015) and Applications
Iova et al. (Iova RPL: The Routing 2016 ☆ × × × × × ☆ × × ★ × × ★
et al., 2016) Standard for the Internet
of Things … Or Is It?
Oliveira and Low-power and Lossy 2016 ☆ × × × × × × × × × ☆ × ×
Vazao ( Networks under
3

Oliveira and mobility: a survey

Vazao, 2016)
Zaho et al. ( A comprehensive study 2017 ☆ ☆ ☆ × × × × ★ ☆ ★ × × ★
Zhao et al., of RPL and P2P-RPL
2017) routing protocols:
Implementation,
challenges and
opportunities
Liu et al. (Liu Performance Analysis of 2017 ☆ ★ ★ × × × × ☆ ✓ ★ × × ☆

Journal of Network and Computer Applications 207 (2022) 103476

et al., 2017) Routing Protocol for Low
Power and Lossy
Networks (RPL) in Large
Scale Networks
Kim et al. (Kim Challenging the IPv6 2017 ★ × × × × × × ☆ ☆ ★ ★ × ★
et al., 2017a) Routing Protocol for
Low-Power and Lossy
Networks (RPL): A
Survey
Triantafyllou Network Protocols, 2018 ☆ × × × × × × ☆ × × ☆ × ☆
et al. ( Schemes, and
Triantafyllou Mechanisms for Internet
et al., 2018) of Things (IoT): Features,
Open Challenges, and
Trends
Witwit and A Comprehensive 2018 ★ × × × × × × × × ★ ★ × ☆
Idrees ( Review for RPL Routing
Witwit and Protocol in Low Power
Idrees, 2018) and Lossy Networks
(continued on next page)
 K.A. Darabkh et al.
Table 1 (continued )
Article Article Title Year Extensive RPL Trickle Routing DODAG Exclusive DODAG Routing Recent RPL Recent Recent Recent Historical Research
specifications algorithm metric/ construction examples local table for applications RPL RPL RPL bibliometric recommendations
constraints comprehensive with repair nodes simulators challenges enhanced analysis
OFs examples different example and root protocols
OFs

Lim, Chansook A Survey on Congestion 2019 ☆ × ★ × × × × × × × ★ × ✓

(Lim, 2019) Control for RPL-Based
Wireless Sensor
Networks
Sobral et al. ( Routing Protocols for 2019 ☆ × ☆ × × × × × × × ✓ × ✓
Sobral et al., Low Power and Lossy
2019) Networks in Internet of
Things Applications
Kharrufa et al. ( RPL-Based Routing 2019 ☆ ☆ ☆ × × × × ★ × ★ ✓ ☆ ☆
Kharrufa Protocols in IoT
et al., 2019) Applications: A Review
Ghaleb et al. ( A Survey of Limitations 2019 ★ ☆ ☆ × × × × ★ ☆ ☆ ✓ × ✓
Ghaleb et al., and Enhancements of the
2019) IPv6 Routing Protocol for
Low-power and Lossy
Networks: A Focus on
Core Operations
Lamaazi and A comprehensive survey 2020 ☆ × ✓ × × × × × × ★ ★ × ☆
Benamar ( on enhancements and
Lamaazi and limitations of the RPL
Benamar, protocol: A focus on the
2020a) objective function
Safaei et al. ( Impacts of Mobility 2020 ☆ × × × × × × × × ★ ★ × ×
4

Safaei et al., Models on RPL-Based

2020) Mobile IoT
Infrastructures: An
Evaluative Comparison
and Survey
Avila et al. ( Security Aspects for Rpl- 2020 × × × × × × × × × ★ ★ ☆ ✓
Avila et al., Based Protocols: A
2020) Systematic Review in IoT
Verma and Security of RPL Based 2020 × × × × × × × × × ★ ★ × ✓

Journal of Network and Computer Applications 207 (2022) 103476

Ranga ( 6LoWPAN Networks in
Verma and the Internet of Things: A
Ranga, Review
2020a)
Rakesh and A Review on Enhanced 2021 ★ × ☆ × × × × × × ★ ★ × ×
Sultana ( Routing Solutions in RPL
Rakesh and Protocol
Sultana,
2021)
This article Recent Advances of 2022 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Routing Protocol for Low
Power and Loosy
Networks over Internet
of Things: A Survey,
Recommendations, and
Future Directions

Note: Symbol (✓) means the side is fully covered in the survey paper, symbol ( × ) means the side is not covered, the symbol (☆) means the side has minor coverage, and the symbol (★) means the aspect is sufficiently
covered.
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

evaluated RPL performance by two methods using two very different in a succinct manner, as well as current research efforts related to each
simulators. They first evaluated the utilization of sink mobile nodes in challenge from 2010 to 2021. In (Triantafyllou et al., 2018), the authors
WSNs to lengthen network lifetime using a WSNet simulator according presented a thorough analysis and classification of all network
to their RPL module. Then they evaluated the use of Power Line communication technologies suitable for the IoT platform while
Communication (PLC) nodes in WSNs using the COntiki Operating focusing more on the valuation and qualification of appropriate network
System (OS) JAva (COOJA) simulator based on their PLC module. Their encapsulation as well as routing protocols. They further examined the
results showed that PRL in the WSNet simulator provides fascinating relationship between IoT network protocols and IoT application devel­
capabilities for managing mobility. Overall, this work is a good refer­ opment. Moreover, they provided a comprehensive layer-based protocol
ence for proper simulators that support the RPL protocol. On the other classification and explained how network protocols are suited and
hand, the authors did not recommend any enhancements to RPL or operated to handle modern IoT requirements and applications. How­
implementation of the RPL protocol with PLC nodes. Tsvetko Tsvetkov ever, they provided a short synopsis of RPL specifications and were
in (Tsvetkov, 2011) presented an overview of the main features of RPL limited to mentioning only five RPL-based protocols out of a wide
and described the protocol’s terminology. Strictly speaking, the author number of existent RPL-enhancements.
discussed the process of constructing the RPL topology and the four Several other surveys have recently been published, which reviewed
different types of PRL control messages used to maintain the topology. some of the new solutions proposed to address the deficiencies of RPL
Additionally, he reviewed RPL metrics. However, apart from the esti­ (Witwit and Idrees, 2018) (Lim, 2019) (Kharrufa et al., 2019) (Ghaleb
mated efforts made in this study, the main challenges and feasible so­ et al., 2019). Among them is the survey paper presented by Witwit and
lutions for implementing RPL, which is one of the main objectives of our Idrees (2018), which addressed the foremost challenges of RPL in LLNs.
survey, were not considered. Al-Fuqaha et al. (2015) provided a general The authors in this study exhibited a comprehensive illustration of the
overview of the IoT, some technical details related to enabling IoT basic characteristics of RPL along with a comprehensive analysis of RPL
technologies, protocols, and applications, and some of the major chal­ operations, such as the construction, repair, and maintenance phases of
lenges of the IoT that have been presented in the literature. Furthermore, DODAG. On top of that, they presented the latest improved versions of
the survey discussed the relationship between the IoT system, big data RPL. The authors also identified open research challenges related to RPL
analytics, fog, and cloud computing. The RPL was partially covered in design and suggested future directions for this routing protocol as they
this study since it is the standard network layer protocol for the IoT. recommend improvement of the OF specification. However, they sum­
They confirmed that the RPL domain is steered towards the highly marized a restricted number of the RPL enhanced versions in a table
appreciated IoT applications, protocols, challenges, and simulation without mentioning any details about them. Also, they did not review
platforms. They also mentioned the significant issues of the IoT, from RPL applications or RPL simulators. In light of the recent challenges of
which most of the RPL challenges stem. However, compared to our RPL, the author in (Lim, 2019) has limited his work to studying the issue
survey, which discusses RPL from all aspects, this survey provides a brief of congestion in RPL. They reviewed the proposed RPL schemes for CC
discussion of RPL. The authors in (Iova et al., 2016) analyzed whether and load balancing and then classified these schemes according to
RPL met the original requirements set by the IETF regarding traffic various criteria: traffic patterns, routing metrics, adopting cross-layer
patterns, mobility, resource heterogeneity, reliability, and scalability. In methods, and exploiting path diversity. They also suggested future
view of this analysis, they highlighted the current trends of IoT and new research directions that include RPL design bearing in mind better-able
requirements that might challenge the future acceptance of RPL as an hardware platforms, examination via actual implementation and ex­
IoT routing protocol. Nevertheless, the study did not evaluate any of the periments, interoperability with Transmission Control Protocol (TCP),
RPL extensions proposed in the scientific literature to overcome its use of multiple RPL instances, as well as cross-layer based design. Sobral
challenges. Oliveira and Vazão (Oliveira and Vazao, 2016) surveyed et al. (2019) performed an extensive survey on routing protocols for
several proposed mobility extensions for RPL and analyzed how the LLNs in IoT applications, focusing on studying the latest solutions to
presented mechanisms affect LLNs requirements. However, the scope of routing issues in these networks. Furthermore, they discussed the
this survey was only limited to RPL extensions intended to deal with strengths and shortcomings of existing routing solutions and stated that
node mobility. Zhao et al. presented an in-depth study of the RPL and the light ad hoc on-demand distance vector-next generation protocol
Point-to-Point RPL (P2P-RPL) protocols regarding their specifications and its developments have arisen as a possible solution, particularly for
and routing approaches (Zhao et al., 2017). They further investigated peer-to-peer communication, although it is less studied than the RPL
the efficacy and flexibility of these two protocols using Network protocol in the current literature. Finally, the authors identified the
Simulator-3 (NS-3). However, this study did not assess the routing remaining open issues and suggested future directions. However, their
overhead of the P2P-RPL protocol nor compared it with that of RPL, in paper is devoted to studying routing solutions for IoT-LLNs and is not
spite of its importance. The review paper by Liu et al. (2017) provided a limited to RPL improvements. In contrast, our work is devoted to an
thorough overview of the main features, metrics, and OFs of RPL. They in-depth study of RPL for the IoT, where we have given it a prominent
also performed a comprehensive analysis of RPL performance with place in this paper. Thus, we investigated all aspects of this protocol, as
OMNeT++ on a large scale. Moreover, they extensively discussed se­ we present later.
curity issues in RPL and future challenges for this protocol. However, the More recently, Kharrufa et al. studied the enhancement of RPL-based
authors reviewed only a few articles related to the OFs of RPL without routing protocols (Kharrufa et al., 2019). This work provided a brief
providing a detailed analysis of the feasibility of the reported studies. illustration of the basic views of RPL. Apart from this, it has covered
Additionally, they have never examined any of the recent solutions some of the common applications that fall under the umbrella of LLNs.
proposed to lessen the shortcomings of RPL regarding downward rout­ Moreover, they draw attention to the contemporary challenges of RPL
ing and routing maintenance. Kim et al. (2017a) presented a survey and how researchers have proposed new strategies to overcome them.
paper that reviews the history of RPL’s research efforts from a variety of They provided limited information about RPL, unlike our survey, and
perspectives from 2012 to 2016. Specifically, they examined and did not mention the simulators used for RPL or future research di­
analyzed 97 scientific research papers studying RPL based on sub-topic rections. Ghaleb et al. presented a thorough survey and detailed
categorization, year of publication, method of evaluation, place of comparative analysis of several routing improvements that have been
publication, and demographics. However, the central focus of this sur­ made to overcome RPL weaknesses (Ghaleb et al., 2019). Specifically,
vey was to report on the success of RPL as a routing protocol. On the they provided an extensive background on the LLNs’ environments,
contrary, our survey mainly focuses on reporting the success of the latest their communication technologies, and their routing requirements. They
solutions that have been introduced in the scientific literature to over­ also addressed RPL’s current challenges regarding its core operations,
come RPL challenges. Additionally, our survey presents the challenges such as routing selection, routing maintenance, optimization

5
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

techniques, and downward routing primitives. Furthermore, the authors Table 1 shows a broad comparison of the 20 highly cited articles
analyzed further the research efforts to address those challenges. Ulti­ mentioned above and shows how our work is distinguished from these
mately, they sensitized researchers regarding the open issues and future earlier ones. The comparison was made on the basis of important RPL-
directions of RPL. Unlike our survey, which covers the majority of en­ related topics covered in these papers, such as extensive RPL specifica­
hancements from 2010 to 2021. In (Lamaazi and Benamar, 2020a), the tion, trickle algorithm details, routing metrics and constraints, OF,
authors proposed a comprehensive overview of the routing metrics and comprehensive examples of DODAG construction, local repair, building
constraints, the standard OFs, and their limitations. Also, they reviewed routing tables between nodes and the DODAG root in both storing and
the approaches that depend on the OF optimizations. This survey paper non-storing modes, recent RPL applications, RPL simulation tools, RPL
is only dedicated to OF challenges and their solutions. The mobility challenges, enhanced RPL protocols, historical bibliometric analysis,
challenge has been thoroughly studied in (Safaei et al., 2020). The au­ and research recommendations. In Table 1, four symbols are used to
thors classified RPL-mobility protocols and compared them according to show how the survey paper has covered the aspect related to a particular
a set of specifications. Furthermore, they explained the motion of the topic. To be more specific, the symbol (✓) is used to designate the aspect
mobile nodes in different protocols in detail by giving advantages and that is fully covered in the survey paper, while the symbol ( × ) is used to
disadvantages of each protocol, a graphical representation of nodes designate the aspect that is not covered. On the other hand, the two
movements, and specific applications related to the suggested protocol. symbols (☆, ★) indicate the degree of coverage of the aspect. For
Finally, they conducted a massive number of simulations to analyze each example, the symbol (★) indicates that the aspect is sufficiently covered,
protocol from different performance perspectives. On the contrary, our while the symbol (☆) indicates that the aspect in the survey has minor
survey addresses the various challenges and briefly evaluates the coverage.
pertinent protocols, whereas their survey focused exclusively on In a nutshell, the main contributions of this survey paper can be
studying mobility on the RPL. summarized as given below:
Since the RPL is vulnerable to a variety of attacks that exhaust node
resources and impair network performance, security was comprehen­ • Considering the main limitations of other surveys presented in
sively studied in (Avila et al., 2020). They investigated 53 RPL solutions Table 1.
between 2011 and 2020 that focused only on the major attacks in the • Making this article a basis for future research by providing the reader
RPL. They reviewed the attacks mitigated and the techniques applied to with sufficient information about the RPL protocol, what has been
overcome these attacks for each studied solution. Another survey that done to enhance RPL functionality, what the current challenges
covered security issues was presented in (Verma and Ranga, 2020a). facing RPL are, what still needs to be developed, and future di­
They conducted a systematic study on different attacks on the RPL, their rections for improving RPL performance.
enhanced proposals, and challenges, as well as open issues and future • Providing a comprehensive overview of RPL specifications, termi­
directions related to the security challenge. Compared with our survey, nology, applications, and simulation platforms, along with giving
the authors of (Avila et al., 2020) and (Verma and Ranga, 2020a) placed informative and brand-new examples covering all of its specifica­
a strong emphasis on RPL security, without considering other issues and tions, makes it very clear to those novices and even expert RPL
provided scant information about RPL specifications. In the end, the researchers.
authors in (Rakesh and Sultana, 2021) reviewed the challenges of the • Describing the basic operation of the RPL protocol, which consists of
RPL protocol, including mobility, multicast communication, RPL OFs, three phases, namely the DODAG construction phase, the DODAG
and energy efficiency, and accordingly, a limited number of proposed detection and repair phase, and the data forwarding phase. This
solutions based on those challenges and their drawbacks. On the other essential information helps researchers to develop and design an
hand, our work extensively addresses almost all the RPL challenges with enhanced version of RPL that avoids problems caused by RPL
the corresponding enhanced solutions, including their techniques, ad­ dysfunction.
vantages, and disadvantages. • Giving a complete summary of the currently available RPL simula­
tion platforms by comparing several simulators based on their main
characteristics, which primarily include RPL support, wide area
1.2. Contributions and impacts networks support, applicable platforms, and programming
languages.
As opposed to surveys in the RPL field, we contemplate giving a • Highlighting the challenges researchers face in developing RPL and
holistic overview and extensive discussions on the essential technolog­ providing a detailed summary of each challenge and what is needed
ical barriers to RPL implementation and their pertinent solutions. to address it based on its distinctive feature.
Generally speaking, this study is conducted to investigate whether the • Presenting a comprehensive review and in-depth study of the
proposed solutions tackled the challenges and downsides that have been cutting-edge RPL-based routing protocols available in recent litera­
highlighted. By virtue of that, this paper is one-of-a-kind in that it ex­ ture (2010–2021), which have been categorized on the basis of their
amines the major aspects of recent efforts to address RPL’s standard primary objectives of providing potential solutions with their limi­
limitations. We really would like to know if the research efforts to solve tations to some of the challenges related to RPL, such as efficient
the inadequacies of RPL have been successful or if more work is needed. routing support, energy consumption, diversification of OFs, QoS,
Then, if so, we direct the researchers as to what they should do with mobility, load-balancing, congestion detection, and cross-layer issue.
regard to this. • Providing a series of tables for each category of these solutions that
Since the standardization of RPL as a first-class IoT routing protocol, summarize the features of the studied RPL-based protocols such as
until now, the number of surveys conducted on various aspects of RPL, their main aims, strategies, strengths, and limitations. These tables
and other related concerns are rare compared to other WSN/IoT routing are intended to help readers quickly obtain more comprehensive
protocols. To the best of our knowledge, there are no earlier works like information about RPL-based solutions without having to read all the
ours. Interestingly, this work pays particular attention to further speci­ published work in this field. Moreover, they provide a starting point
fications on RPL, other recent and relevant RPL issues, and compre­ for researchers to make further improvements to existing RPL rout­
hensive examples of DODAG construction and local repair, which ing solutions.
constitute one of the primary contributions of this work. Intriguingly,

6
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 2
List of abbreviations and their definitions.
Abbreviation Definition Abbreviation Definition Abbreviation Definition

6LoWPAN IPv6 over Low-Power Wireless HC Hop Count OF Objective Function

Personal Area Networks
ALBER Adaptation Layer Between BLE and ICT Information and Communication OF-ECF Objective Function based on ETX, Consumed
RPL Technology Energy, and Forwarding Delay
AMI Advanced Metering Infrastructure IEEE Institute of Electrical and Electronics OF-FL Objective Function Fuzzy Logic
Engineers
A-LLNs Agricultural LLNs ICMPv6 Internet Control Message Protocol OF0 Objective Function Zero
version 6
ACO Ant Colony Optimization IETF Internet Engineering Task Force OFQS OF with QoS differentiation
BPCC Back Pressure Congestion Control IoT Internet of Things OHCA Optimization-based Hybrid Congestion
Alleviation
BRPL Backpressure RPL IPv6 Internet Protocol version 6 OS Operating System
BMRF Bidirectional SMRF KP-RPL Kalman Positioning RPL O2F Optimization-based OF
BFCC Bird Flocking Congestion Control L-RPL Laplacian-RPL PAOF Parent-Aware Objective Function
BLE Bluetooth Low Energy LA-OF Learning Automata OF PHY Physical
Co-RPL Cognitive RPL L2AM Lifetime and Latency Aggregately P2MP Point to Multipoint
Metric
CA-RPL Congestion Avoidance Multipath- RPL LLQE Lightweight Link Quality Estimation P2P Point to Point
CC Congestion Control LC Link Color P2P-RPL Point-to-Point RPL
CT Connectivity Timer LQ Link Quality PLC Power Line Communication
CoAP Constrained Application Protocol LQL Link Quality Level PC-RPL Power-Controlled-RPL
CL-RPL Context-aware and Load-balancing LB-RPL Load- Balance - RPL Prf Preference
RPL
COOJA COntiki OS JAva Simulator LBER-RPL Load Balanced – Efficient Routing RPL PMFR Proactive Multicast Forwarding with RPL
CSMA Carrier Sense Multiple Access LOA-RPL Lion Optimization Algorithm RPL QCOF QoS and Congestion-Aware OF
DMC DAG Metric Container LoBaPS Load Balancing Parent Selection for QoS Quality of Service
RPL
DMR DAG Multipath Routing LLNs Low Power and Lossy Networks QoS_RPL Quality of Service-RPL
DQCA-OF Delivery Quality and Context-Aware LPP Low-Power Probing QU-RPL Queue Utilization- RPL
Objective Functions
DAO Destination Advertisement Object LR-WPANs Low-Rate Wireless Personal Area RDC Radio Duty Cycling
Networks
DAO-ACK Destination Advertisement Object MTU Maximum Transfer Unit RSSI Received Signal Strength Indication
Acknowledgment
DTSN Destination Advertisement Trigger MAC Media Access Control REFER Reliable and Energy-Efficient RPL
Sequence Number
DODAG Destination-Oriented Directed Acyclic MRHOF Minimum Rank with Hysteresis RT Reply Timer
Graph Objective Function
DAG Directed Acyclic Graph MSN Mobile Sensor Network ROLL Routing Over Low-power and Lossy Networks
DT-RPL Diverse Traffic-RPL MARPL Mobility Aware RPL RPL Routing Protocol for LLNs
DIO DODAG Information Object MDT Mobility Detection Timer RPL-FL RPL based on the Flexible Trickle Algorithm
DIS DODAG Information Solicitation ME-RPL Mobility Enhanced-RPL SCAOF Scalable Context-Aware Objective Function
DCRL-RPL Dual context-based Routing and Load MLEq-RPL Multiple gateway Load balancing SOF-QoS Service-aware OF with QoS
Balancing in RPL Scheme for Equilibrium
DCCC6 Duty Cycle Aware Congestion Control MM-RPL Mobility Management for RPL SSCF Sink-to-Sink Coordination Framework
D-OF Dynamic Objective-Function mRPL mobility-RPL SG Smart Grid
D-RPL Dynamic RPL MOORA Multi-Objective Optimization on the SH Smart Home
basis of Ratio Analysis
ELITE Elaborated Cross-Layer RPL OF to MOP Mode of Operation SMS Smart Manufacturing System
achieve Energy Efficiency
EAOF Energy-Aware Objective Function MOF MOORA OF SMRF Stateless Multicast RPL Forwarding
EMA-RPL Energy-efficient and Mobility-Aware mQFQS multi-objective metric OFQS SPR Strobe per Packet Ratio
routing protocol
EL-RPL Energy-Efficient Load-Balanced RPL MRRDþ Multiple, RSSI, Rank and Dynamic T-OF Time OF
ER-RPL Energy-efficient Region-based Routing MP2P Multipoint to Point TSCH Time-Slotted Channel Hopping
Protocol
ESMRF Enhanced SMRF NETSTACK Network Protocol Stack TSCH-RPL Time-Slotted Channel Hopping - RPL
ER-RPL Enhancing the Reliability of RPL ND Neighbor Discovery TCP Transmission Control Protocol
ELT Expected Lifetime metric NLQ Neighbor Link Quality UDP User Datagram Protocol
EMT Expected Multicast Transmissions NS-3 Network Simulator-3 UML Unified Modeling Language
ETX Expected Transmission Count NDR-Ack New DODAG Acknowledgment WSNs Wireless Sensor Networks
message
EKF-MRPL Extended Kalman Filter for Mobile NDR New DODAG Request message W-SPH Wireless Shortest Path Heuristic
RPL
FIS Fuzzy Inference System NE Node Energy WG Working Group
GTCCF Game Theory-based Congestion NSA Node State and Attribute ZoA Zone of Attraction
Control Framework
GTCC Game Theory-based Congestion NOCS Number of Children and Siblings ZoR Zone of Repulsion
Control
GRA Grey Relational Analysis OCP Objective Code Point

7
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

• Concluding the RPL challenges and enhanced solutions sections with by the IETF as RFC 6550 in March 2012. In fact, the RPL is designed with
the authors’ insights, which are a summary of the findings we the goal of meeting the routing requirements defined by the IETF ROLL
observed after analyzing a substantial amount of research. WG in 2009 for four LLNs application areas: home automation LLNs
• Conducting a historical bibliometric analysis of publications on RPL (RFC 5826) (Brandt et al.), building automation LLNs (RFC 5867)
challenges addressed from 2010 to 2021. The analysis mainly covers (Martocci et al., 2010), urban LLNs (RFC 5548) (Dohler et al.), and in­
the number of scientific articles published annually for each chal­ dustrial LLNs (RFC 5673) (Pister et al., 2009). Furthermore, it is
lenge. Also, this bibliometric analysis was carried out using 509 ac­ designed to be capable of operating across a diverse range of data link
ademic works, which were identified by keyword searches layers, including those that are restricted, potentially lossy, or typically
performed on Google Scholar, Scopus, and IEEE Xplore databases. used with devices with inadequate resources, such as transportation,
Based on our analysis, we make recommendations and identify home, and industrial applications.
eleven substantive directions for future research that must be RPL’s ability to quickly build network routes and effectively adapt to
considered while designing new RPL-based protocols. As a result, network topology makes it a perfect option for 6LoWPAN devices
this work will be a useful reference for all RPL designers and (Prasad and Babu, 2017). RPL also provides a set of other features,
developers. including self-configuration, which incorporates the use of IPv6, which
enables the operation of ND for the establishment of network routes for
1.3. Paper organization data transmission. When nodes are added or deleted in the RPL network
dynamic topology or a node malfunctions, the auto-healing (repairing)
The remainder of this survey paper is systematized as follows: Sec­ capability comes in handy (Yang et al., 2021). In RPL, several border
tion II provides a comprehensive overview of the RPL protocol’s basic routers offer high stability and load balancing. Table 3 summarizes other
features. Section III provides a full demonstration of RPL’s fundamental key features of the RPL protocol.
operations. The most popular RPL applications are summarized in Sec­
tion IV. Section V covers the simulation platforms that RPL supports in 2.1. RPL terminology
detail. The RPL protocol faces several challenges, which are discussed in
Section VI. The state-of-the-art RPL-based solutions that address This subsection provides a brief description of the RPL terminology
contemporary challenges are provided in Section VII. Section VIII pre­ used in this study and is intended to help readers better understand the
sents the future directions and recommendations, which are concluded concepts presented in the following sections. The following terminology
by the authors. Finally, Section IX brings the survey to a close. is adapted from (Winter et al., 2012):

• DAG – A directed acyclic graph is a digraph (or directed graph) in

1.4. List of abbreviations and acronym
which all edges are oriented so that there are no directed cycles. All
of the edges are included in paths oriented to and terminated at one
To make the reader easily search for unfamiliar abbreviations, we
or more root nodes.
provide an alphabetical list of all abbreviations used in this paper and
• DAG root – It is a node with no outgoing edge. Since the graph is
their corresponding definitions in Table 2.
acyclic, accordingly, all DAGs should have one or more DAG roots
and all the paths should end at the DAG root.
2. RPL overview • DODAG – A Destination-Oriented DAG is a DAG rooted in one
destination, for example, at one DAG root (DODAG root) without
The impetuses for standardizing RPL, terminology, types of RPL outgoing edges.
traffic patterns, routing metrics, OFs, control messages, and the trickle • DODAG root – It is the DAG root of DODAG that can also operate as a
algorithm, are covered in this section. RPL is a loop-free distance vector DODAG edge router. In specific, it can aggregate routes into DODAG
routing protocol developed by the ROLL WG for wireless networks that and can redistribute DODAG routes to other routing protocols.
have low power consumption and are generally prone to packet loss. It • RPL Instance – A RPL instance is a collection of single or multiple
specifically uses a proactive approach for LLNs (Kim et al., 2017a), DODAGs under the same RPLInstanceID. Because each RPL instance
which means that it finds and maintains all routes across the network operates irrespective of the others, a RPL node can only pertain to a
even if they are not being used. RPL operates on the IEEE 802.15.4 single DODAG in a RPL instance.
standard, optimized for multi-hop and multi-to-one communication, but
as well supports one-to-one messaging. This protocol was standardized

Table 3
Key features of RPL routing protocol.
Feature Description Feature Description

Publication data March 2012 Control messages DIS, DIO, DAO, and DAO-ACK
Type Proactive Neighbor discovery IPV6 ND mechanisms
Traffic pattern MP2P/P2MP/P2P Mobility Static - Mobile
IPv6 support Yes Transmission Unicast and multicast
Topology Hierarchical/Flat Scalability High
Memory usage Low Message update Trickle timer
MOP Storing and non-storing Local repair Yes
Algorithm Source routing/Distance vector Energy usage Low
Mechanism - Using Energy-aware metrics IoT challenges solved - Local and global repairs
- Multipath routing - Low Energy usage
- Using trickle algorithm - Support mobility
- High scalability
- Low memory usage
Advantages - Loop avoidance and detection Disadvantages - No security
- Timer management

8
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Fig. 2. Types of traffic flows supported by RPL.

• Upward – Upward route indicates the direction from the leaf nodes towards the DODAG root (preferred parent). The rank of a DODAG
towards the DODAG roots. parent is lower than that of a node.
• Downward – Downward route indicates the direction from the • DAO parent – A node to which the unicast DAO messages are sent.
DODAG roots towards the leaf nodes. • DAO parent set – The collection of DAO parents for a node.
• Rank – The rank of a node determines the distance of a node in
relation to other nodes with regard to the DODAG root. The rank 2.2. Traffic flows supported by RPL
decreases strictly in the upward direction and the other way around
the downward direction. The rank’s is obtained from the OF. This subsection informs readers about the different types of traffic
• OF – An OF describes how nodes along a DODAG choose their paths patterns investigated in RPL. From a technical standpoint, the RPL
by determining their parents. Further to that, it allows nodes to turn protocol supports three basic traffic flows:
a set of metrics and constraints into a numerical value known as
“rank.” Thus, the OF is responsible for constructing the DODAG. • MP2P traffic: MP2P is a predominant traffic flow in numerous LLN
• OCP – An objective code point is a unique identifier that specifies applications. The destinations for MP2P flows are specific nodes that
which OF is supported by the DODAG. have some significance to the application, for example, giving con­
• DIS – DODAG Information Solicitation (please see Subsection E for nectivity to the greater Internet or the underlying private IP network.
its usage). The MP2P traffic is supported by RPL by allowing access to MP2P
• DIO –DODAG Information Object (please see Subsection E for its destinations via DODAG roots. Fig. 2(a) shows the direction of
usage). messages in MP2P, in which all router nodes in the DODAG send
• DAO –Destination Advertisement Object (please see Subsection E for messages to the DODAG root, which ultimately creates an upward
its usage). route (Kaviani and Soltanaghaei, 2022).
• DAO-ACK –Destination Advertisement Object Acknowledgment • P2MP traffic: P2MP is a traffic pattern needed for many LLN ap­
(please see Subsection E for its usage). plications. RPL supports P2MP traffic by utilizing a destination
• RPLInstanceID – The RPLInstanceID uniquely identifies the declaration mechanism, which provides routes downward to desti­
network. All DODAGs with the same RPLInstanceID have the same nations (prefixes, addresses, or multicast groups) and away from the
OF. roots. This mechanism may update routing tables when the DODAG
• DODAGID – A DODAGID is the identifier for the DODAG root that is topology changes. Fig. 2(b) demonstrates the P2MP message direc­
unique within the domain of a RPL instance. A DODAG is recognized tion, in which the DODAG root sends messages to all other router
by the pair (RPLInstanceID, DODAGID). nodes, forming a downward route (Seyfollahi and Ghaffari, 2021).
• DODAG Version Number – A serial counter that the root can • P2P traffic: The fundamental construction for P2P traffic is sup­
increment to release a new version of a DODAG whenever there is a ported by RPL. To achieve this purpose, the DODAG root must have
global repair. The DODAG Version is determined by a triple the ability to forward packets at the point where the source is routed
(RPLInstanceID, DODAGID, and DODAG Version Number). to the destination. Each node within the DODAG might as well have a
• Goal – The goal is to determine the application requirements that are routing table to the destination. The packet travels from the root
specified outside the bounds of the RPL. The DODAG root should until it finds an ancestor who knows how to get to the destination.
know this goal to decide if the goal can be achieved or not. A typical When the DODAG is in the non-storing mode (i.e., the nodes do not
goal is to construct a DODAG that adheres to a particular OF while store paths), the DOADG root is the common ancestor. Fig. 2(c)
maintaining connectivity to a set of nodes. The goal, for example, is points out the P2P traffic, in which a node sends packets to the other
to implement an OF that minimizes a metric (i.e., HC). (non-root) node of DODAG. Hence, both the upward and downward
• Grounded – A DODAG is known to be grounded when the DODAG routes may be required (Kim and Paek, 2020).
root achieves its goal.
• Floating - If a DODAG is not grounded, it is known as floating. 2.3. Routing metrics and constraints used by RPL
Although a floating DODAG is not achieving the goal, it can grant
connectivity to other nodes in the DODAG. In this subsection, we review the basic metrics and constraints that
• DODAG parent – Each node within a DODAG contains a set of are used in RPL. Routing protocols use routing metrics to figure out
parents, one of which is a direct successor of the node on a path which routes are the shortest. In LLNs, both node and link metrics are

9
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 4
Summary table for equations’ parameters.
Parameter Definition Equation number

R(n) Rank of node 1

R(P) Rank of preferred parent 1
Ri Rank increase 1
Rf Rank factor 2
Sr Stretch of rank 2
MHRI Min Hop Rank Increase 2
TIRnoden True Integer Rank of a node n 3
PPack Probability of data packet is successfully received by the destination/neighbor 4
PACK Probability of the acknowledgment packet was successfully received 4

required, which may be either static or dynamic metrics. Indeed, in • LQL Object: The LQL object is utilized to estimate link reliability
theory, it is very difficult to assign a single or composite metric so that utilizing a discrete value from 0 to 7. A value of 0 demonstrates that
the requirements of all applications of the RPL protocol are met. Addi­ the LQL is unknown, and a value of 1 is the highest LQ. The LQL can
tionally, RPL authorizes the use of routing constraints where those be utilized as a constraint or a metric. When utilized as a metric, only
constraints are implemented in both links and nodes. RPL enables a node the LQL metric can be recorded (Iova et al., 2013).
or link that does not meet the requirements of a certain constraint to be Here are other examples of node metric/constraint objects (Barthel
removed from the candidate neighbor set. Meanwhile, a constrained- et al., 2012) (Kamgueu et al., 2012) (Capone et al., 2014a):
shortest path is guaranteed (Liu et al., 2017). In (Barthel et al., 2012), • HC Object: The HC object is utilized to inform the number of nodes
the authors provide a complete set of supported constraints and metrics tracked along a route. The HC can be utilized as a metric or a
for both nodes and links, which can help researchers find the appro­ constraint. When utilized as a constraint, the DODAG root demon­
priate OF that is used to compute the node’s rank. Here are some ex­ strates the greatest number of hops a path can cross. At the point
amples of link metric and constraint objects: when this number is reached, no further node can join this route.
However, when utilized as a metric, the total number of hops along
• ETX Object: The ETX metric represents the expected number of the route to the destination is counted.
transmissions that a node in the network will make to effectively • NSA Object: The NSA object is utilized to provide information about
deliver a packet to the destination. Details of this metric are node characteristics, including workload and traffic aggregation.
described in the next subsection. However, an ETX object can be Moreover, it can be utilized to reroute traffic when the router is stuck
utilized as a path metric or a constraint. For instance, an ETX object in a congestion situation.
can be utilized as an additive aggregate metric as the value is • NE Object: The NE object is utilized to provide information about
refreshed along the path to reveal the path quality (Badach, 2018) the node’s energy to avoid choosing a node with low remaining
(Solapure and Kenchannavar, 2020). In another situation, it might be energy as a router and can also be used as a metric or a constraint. NE
necessary that the ETX does not surpass a certain threshold. The indicates the energy consumed by nodes throughout not only their
common header of an ETX object, in this case, indicates that the communications (i.e., messages’ exchanges) in the network but also
value is bound to a constraint. their central processing unit computations (or extra space).
• Throughput Object: Throughput is the data rate of the link, which is
generally measured in bytes per second. A node in an LLN must
notify other nodes if there is a change in the throughput to enable its 2.4. Objective function (OF)
links to handle that change.
• Latency Object: The latency can be used as a path metric or a In this subsection, we present the main concepts about OFs, along
constraint. For instance, the latency object can be used as an additive with the main two OFs in RPL. The OF specifies how RPL nodes within a
aggregate metric as the value is refreshed along the path for RPL instance define and optimize routes. The OF is determined by an
reflecting the path latency. In another instance, one might wish for OCP under the DIO configuration option. In fact, the OF specifies how
the latency to not surpass a threshold. In this situation, the latency RPL nodes translate one or more of the metrics and constraints defined
object’s common header points out that the value provided is related in RFC 6551 (Barthel et al., 2012) into a value named rank, which ap­
to a constraint. proximates the distance of the node from the DODAG root. The OF also
• LC Object: The LC object is a 10-bit link administrative constraint determines how nodes are selected for parents. Each RPL instance is
(which can be either dynamic or static) that is used to attract or avoid associated with a particular OF. The RPL standard defined in RFC 6550
specific links for specific types of traffic. The LC object may be uti­ (Winter et al., 2012) does not recommend any particular OF. Rather,
lized either as a constraint or as a metric. When utilized as a metric, there are separate specifications specifying OFs. To date, only two
only the LC object can be recorded. For instance, a DAG might fundamental OFs have been standardized by the IETF ROLL, the OF0 and
require LCs to be recorded for all traversed link. The color of a link is the MRHOF. Intriguingly, the definitions of parameters used in the
determined by a series of bit values. A counter is used to compress the equations of this subsection are summarized in Table 4.
information in which the number of links is reported for each LC
(Barthel et al., 2012) (Alsukayti and Alreshoodi, 2019) (Pancaroglu • OF0
and Sen, 2021). When utilized as a constraint, an LC object can be
inserted into a DMC to point out that links of the specified color must The OF0 is defined in RFC 6552 (Thubert, 2012) and is intended to
be included or excepted from the computed path. find the closest DODAG root that provides adequate connectivity,
although there is no certainty that the path will be optimized based on a
specific metric. The OF0 defines a preferred parent as well as a feasible

10
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

To be further clarified, the ETX is generally considered a metric that

quantifies how good the wireless connection is between two nodes. The
LQ × NLQ is used to calculate the overall LQ for a round-trip between
two nodes. For example, let us suppose that the transmit LQ is 75% and
the received NLQ is 76%, then the round-trip probability is 57%. This
means that 57 packets out of 100 completed a round trip. Consequently,
we can estimate that the expected number of transmissions required to
successfully complete the transmission of a packet over a link, ETX, is 1/
0.57 = 1.7544 (Darabkh et al., 2022b). Nevertheless, if we assume that
the packet reaches a destination in 4 hops, then there are exactly three
Fig. 3. RPL control message structure. possible values of ETX (i.e., ETX1 , ETX2 , and ETX3 ). Thus, the ETX for the
∑3
entire route ETXTOTAL is calculated as ETXk .
k=1

backup successor if available. Normally, all upward traffic is routed 2.5. ICMPv6 RPL control messages
through the preferred parent without any attempt to achieve any load
balancing. R(n) is calculated by adding the rank of a chosen preferred In this subsection, we go over the types of RPL control messages that
parent R(P) to a normalized scalar, Ri .Concretely, R(n) and Ri are let data be transmitted between nodes in the DODAG.
calculated by the formulas given in Equations (1) and (2), respectively, The RPL control message is a new ICMPv6 informative message of
where Rf and Sr are configurable values. SP is an intermediate compu­ Type 155, which has been confirmed by the Internet assigned numbers
tation that depends on the properties of the link with a particular authority. It is defined by a code and consists of a rule depending on the
neighbor. MHRI represents the minimum rank increment between a code plus a series of options. The structure of the RPL control message is
node and any of its DODAG parents whereas its default value is 256. shown in Fig. 3, consisting of an ICMPv6 header and a message body.
When comparing rank to loop detection or to determine parent re­ The ICMPv6 header contains three fields: Type, Code, and Checksum.
lationships, the integer part of the rank must be used. TIRnoden is calcu­ The Code field specifies the type of RPL control message while the
lated as shown in Equation (3) (Winter et al., 2012) (Darabkh et al., Checksum field is used for RPL security operations and is set to 0. On the
2022b). other hand, the message body consists of a message base and perhaps a
number of options. The node drops any received RPL message that has
R(n) = R(P) + Ri , (1) an unknown code field without any further processing (Charles and
( ) Kalavathi, 2022). In order to maintain the topology and exchange in­
Ri = Rf * Sp + Sr *MHRI , (2) formation, RPL utilizes four types of control messages, which are
⌊ ⌋ described below.
R(n)
TIRnode n = . (3) • DIS: The DIS message (Code field = 0 ×00) can be utilized to
MHRI request a DIO from the RPL node. Strictly speaking, the DIS message
induces the sending of DIO because the node, which wants to take part in
• MRHOF the DODAG, transmits a DIS message to the root, which then transmits a
DIO message containing all of the DODAG’s information. On the top of
The MRHOF is defined in RFC 6719 (Gnawali and Levis, 2012) and is that, it can be utilized by a node to explore its neighborhood for adjacent
primarily designed to find paths with the minimum path cost while DODAGs, similar to using a Router Solicitation as defined in IPv6 ND
avoiding excessive churn of the network topology. It does this through (Narten et al., 2007).
the use of two mechanisms. It starts with finding the minimum cost path • DIO: The DIO message (Code field = 0 ×01) is the most important
(i.e., the path with a minimum rank). Second, it converts to this mini­ control message in RPL and is used to find proper upward routes, which
mum rank path just in the case it is shorter (in respect of path cost) than are discussed later. It carries information that permits a node to discover
the present path in at least a certain limit. This second mechanism is a RPL instance, find out its configuration parameters, including RPLIn­
termed “hysteresis”. MRHOF can be utilized with any additive metric stanceID, DODAGID, OCP, Version Number, routing metrics, and cur­
defined in (Barthel et al., 2012) as long as the routing intends to mini­ rent rank, and choose a DODAG parent set, as well as maintain the
mize the specified routing metric. The metric used by the MRHOF is DODAG. It should be emphasized that a DODAG root originates and
defined through the metrics advertised in the DMC option present in the broadcasts DIO to neighbors using the trickle algorithm, which is dis­
DIO messages. For example, using MRHOF with a latency metric permits cussed in more detail in the next subsection (Arslan et al., 2020). The
RPL to find stable paths with minimum latency from nodes to the structure of the DIO Base Object is illustrated in Fig. 4. The main DIO
DODAG root. Using MRHOF with the ETX metric permits RPL to find Base Object fields are described below.
stable paths with minimum ETX from nodes to root in a DAG instance
(Gupta et al., 2021). If there is no metric in the DMC option in the DIO
messages, the MRHOF defaults to utilizing ETX to calculate the rank. As
mentioned earlier, the ETX metric indicates how many transmissions a
node expects to make to a destination with the intention of successfully
delivering a packet. Specifically, the ETX is calculated as shown in
Equation 4. In this Equation, PPack , which points to NLQ, is the proba­
bility that a data packet is successfully received by the destina­
tion/neighbor, and PACK , which points to LQ of a node, is the probability
that the acknowledgment packet was successfully received in return as
well (Darabkh et al., 2022b).
1 Fig. 4. DIO Base Object structure.
ETX = (4)
PPACK × pACK

11
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

1) RPLInstanceID (8-bit): It is a field set by the DODAG root to specify transmission rate of controlling messages (i.e., routing control traffic).
which RPL instance the DODAG belongs to. This can be done in particular by implementing a consistency model to
2) Version Number (8-bit): It is an unsigned integer assigned to the check whether RPL nodes contain outdated routing information. If this is
DODAG Version Number by the DODAG root. This field specifies the case, the trickle algorithm quickly schedules sending a DIO message
the DODAG Version Number, which is usually incremented on to update this information. Otherwise, it drastically reduces the trans­
each update of network information (i.e., global repair is trig­ mission rate of DIO messages. More precisely, each RPL node utilizes a
gered) and helps keep all nodes in sync with new updates. trickle timer to schedule the transmission of DIO messages at a certain
3) Rank (16-bit): It is an unsigned integer signifying the node’s rank time t, which is randomly selected from the trickle interval. The trans­
issuing the DIO message. mission of a scheduled DIO message depends on trickle parameters,
4) Grounded (G) (1-bit): It is a flag specifying whether or not the variables, and a set of rules. The trickle timer, as mentioned in (Levis
DODAG advertised can meet the application-defined goal. The et al., 2011), runs for a specified interval and contains the following
DODAG is grounded if the flag is 1. The DODAG is floating if the three configuration parameters:
flag is 0.
5) MOP (3-bit): It specifies the RPL instance’s mode of operation as • Imin : The minimum interval size specified in seconds or milliseconds.
administratively provided and dispersed by the DODAG root. Imin learned as 2DIOIntervalMin from the DIO message, where
Each node joining the DODAG should be capable of honoring the DIOIntervalMin is advertised along the DODAG in a DIO message by
MOP for the sake of fully participating as a router; otherwise, it the DODAG root (Winter et al., 2012).
should only join as a leaf node. Four MOPs are defined in RFC • Imax : The maximum interval size described as the number of times Imin
6550 (Winter et al., 2012): MOP 0 (No downward routes main­ is doubled, i.e., Imax = Imin × 2Idoubling , where Idoubling is used to deter­
tained by RPL), MOP 1 (Non-storing MOP), MOP 2 (Storing MOP mine how many times Imin can be doubled before sustaining a con­
with no multicast support), and MOP 3 (Storing MOP with mul­ stant multicast rate. Imax learned as DIOIntervalDoublings from the
ticast support). DIO message, where DIOIntervalDoublings is advertised along the
6) DODAG Prf (3-bit): It is an unsigned integer specifying how much DODAG in a DIO message by the DODAG root.
this DODAG root is preferred over the other DODAG roots in the • k: The redundancy constant, where k is a natural number greater
instance. The Prf ranges from 0x00 (the least preferred) as the than zero. k learned as DIORedundancyConstant from the DIO
default value to 0x07 (the most preferred). message.
7) DTSN (8-bit): Each node that sends a DIO message sets this un­
signed integer field. As part of the procedure, the DTSN flag is In addition, the trickle algorithm maintains the following three
required to maintain downward routes. variables:
8) Flags (8-bit): It is an unused field and is reserved for flags. The
sender node must set it to 0, while the receiver must neglect it. • I: The size of the current interval.
9) Reserved (8-bit): It is an unused field. The sender node must set it • t: A time within the current interval.
to 0, while the receiver must neglect it. • c: A counter used to count the consistent messages received in the
10) DODAGID (128-bit): This field is set by a DODAG root. The current interval.
DODAGID uniquely identifies the DODAG and must be a routable
IPv6 address that belongs to the DODAG root. It is worth stating that the following steps refer to the authors’
perspective, and particularly we came up with them after reviewing and
• DAO: The DAO message (Code field = 0 ×02) is utilized by RPL understanding a substantial number of articles in this realm. Accord­
nodes to disseminate destination information upward along DODAG for ingly, the trickle algorithm has the following six rules that describe the
the purpose of establishing and maintaining downward routes. RPL behavior of each RPL node:
specifies two dissimilar modes of downward traffic: storing and non-
storing. In storing mode, DAO messages are unicast to the parent(s) 1. When the algorithm starts executing, it sets the size of the current
selected by the child. In non-storing mode, DAO messages are unicast to interval, I, to a value in the range [Imin , Imax ]. Then the algorithm
the DODAG root. These modes are discussed in more detail in Section III starts in the first interval.
(Cao and Muqing, 2018). 2. When the interval starts, the trickle algorithm resets the counter c to
• DAO-ACK: The DAO-ACK message (Code field = 0 ×03) is a uni­ 0 and sets the value of t to a random value in the interval, selected
cast packet issued by a DIO receiver (i.e., a DODAG root or DAO parent) from the range [I /2, I).
in response to a received unicast DAO message (Kim and Paek, 2020). 3. Whenever trickle hears a “consistent” transmission, it increments the
c counter. Strictly speaking, a RPL node keeps track of DIO messages
2.6. Trickle algorithm received from neighbors by incrementing the c counter each time it
receives a consistent DIO message.
This subsection details the trickle algorithm and how it is exploited 4. At time t, if the value of the counter c is less than the DIO redundancy
in the RPL protocol. Trickle is an efficient self-organizing algorithm constant (k = DIORedundancyConstant), the RPL node transmits the
designed primarily to rapidly propagate code updates in WSNs while scheduled DIO message. In other cases, DIO transmission is
imposing very low maintenance overhead. The trickle algorithm ach­ suppressed.
ieves this using a policy of “polite gossip”, in which nodes periodically 5. When interval I expires (i.e., t > I), trickle doubles the length of the
broadcast a summary of the code to all of their neighboring nodes, but interval. If the length of this new interval is longer than Imax , trickle
remain silent if they have lately heard a summary matching theirs. When sets the length of the interval I to Imax .
a node hears a summary that is older than its own, it broadcasts a code 6. Whenever trickle hears an “inconsistent” transmission and I is larger
update. Rather than flooding the network with packets, the trickle al­ than Imin , trickle resets the RPL trickle timer by setting the I to Imin .
gorithm controls the transmission rate such that each node hears just a Then the trickle starts a new interval just like in step 2.
little trickle of packets, enough to stay up to date (Levis et al., 2004). In
fact, the trickle has proven to be a powerful method that can be used to There are many different inconsistent events in the RPL that cause
solve a wide variety of protocol design issues, such as control traffic the trickle to reset its timer. This includes joining a node to a new
timing, route discovery, and multicast propagation (Levis et al., 2011). DODAG Version, receiving a node to a multicast DIS message, receiving
Specifically, this algorithm is used in the RPL protocol to control the

12
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

root and leaf nodes. They are used to forward and generate traffic and
may advertise structure information to their neighbors. Unlike root
nodes, they do not have the potential to construct a new DODAG but are
associated with existing devices. Leaf nodes, on the other hand, refer to
the end-devices located at the edges of the DODAG topology. They are
able to generate data traffic, but they are not able to forward data traffic
to a terminal device. Fig. 5 illustrates a simple example of a DODAG
topology. The entire RPL protocol strives to build, maintain, and
improve DODAGs. In a DODAG, the nodes are aware of their parents, but
they are unaware of the children associated with them. Furthermore,
each RPL node maintains one route to the DODAG root, as well as a
preferred parent, to continue construction a faster route to increase
network performance.
In general, the RPL operates in three phases: the DODAG construc­
tion phase, the DODAG detection and repair phase, and the data for­
warding phase. Each of these three phases is explained in more detail in
the following subsections.

3.1. DODAG construction phase

Fig. 5. An example of a DODAG topology.
The RPL constructs a DODAG by creating bi-directional routes, i.e.,
a node to a DIO message with a newer Version Number, routing loop upward and downward routes. This subsection describes how RPL finds
detection, changing the node rank, etc. On the other hand, a DIO mes­ and maintains upward and downward routes and what type of control
sage that does not cause any change in the recipient’s preferred parent, messages are used to create these routes that lead to the construction of
rank, or parent set is considered consistent in regard to the trickle timer. the DODAG.
Choosing configuration parameter values for the trickle timer is critical
because it will directly affect network performance and reliability, • Upward Route
especially in large scale networks. For example, the redundancy con­
stant k must be chosen carefully to avoid mismatched k values between The steps of the DODAG construction in upward route are summa­
all nodes, as this may result in uneven transmission load and thus energy rized as follows (Darabkh et al., 2022b) (Mohamed et al., 2021):
exhaustion of some of these nodes, particularly in a low-power network.
Moreover, the constant k should not be set to infinity as it may cause Step 1: The DODAG construction process is initiated by the DODAG
congestion in dense networks (Levis et al., 2011). The Imin should also be root, which periodically broadcasts a DIO message to all its neigh­
set to be at least 2–3 times longer than the time it takes to send k packets boring nodes. This message conveys important parameters for
to avoid congestion and massive packet loss. In addition, high intervals building and maintaining a DODAG topology. These parameters
result in lower response to network inconsistencies, including those include DODAGID, RPLInstanceID, Version Number, rank, MOP,
caused by mobile nodes, whereas lower intervals mean more overhead, metrics, DTSN, and OCP (used for OF).
resulting in a shorter network lifespan (Kharrufa et al., 2017) (Vasseur Step 2: Every time a RPL node receives a DIO message, it must first
and Dunkels, 2010). It is worth noting that only a few research papers decide whether or not to accept this message for further processing.
have focused on trickle algorithm optimization (Lamaazi and Benamar, Concretely, if a received DIO message does not satisfy some criteria
2020b). defined by the RPL (e.g., a malformed DIO message), it is considered
ineligible for further processing and must therefore be discarded by
3. RPL operations the node. Otherwise, the node should process the DIO message.
Step 3: Once a node receives a DIO message for the first time and
Before getting started, it is worth mentioning that the information decides to join the DODAG, it adds the DIO message sender address
available in this section is extracted from (Winter et al., 2012), where it to its parent list, and then calculates its rank based on the OF defined
is detailed here in a new and very professional manner, in addition to in the OCP field, thus its rank is higher than the rank of each parent.
providing informative examples for clarity purposes to be very attractive Then, this node forwards the DIO message containing the updated
for those beginners and even expert RPL researchers. This section pro­ rank information to other nodes in a multicast fashion. It also selects
vides an entire RPL operations, including the upward and downward the most preferred parent from its parents’ list as the default node by
routes, loop avoidance and detection, route repair, and data forwarding. which traffic is redirected inward.
The RPL organizes the network topology as a DAG, which is divided Step 4: In another case, when a node that has already joined a
into one or more DODAGs, one for each root. The DODAG is a tree-like DODAG receives another DIO message, it must process the DIO
routing structure in which nodes are linked to a root node, called the message by calculating its new rank and then comparing it with its
LLN border router or sink node, and has one path from each single leaf previous one. If the new node’s rank is lower than its previous one,
node to the root node (Winter et al., 2012). To construct the DODAG, the the node will add the DIO sender to its parent list and choose it as its
RPL protocol uses the OFs that determine the most appropriate routes in new preferred parent instead of the previous one. The node then
the network. General speaking, there are mainly three different types of updates the rank in the DIO message and broadcasts the latest
nodes in a RPL network (Tsvetkov, 2011): root nodes, router nodes, and version of this message to all of its neighboring nodes.
leaf nodes. Root nodes are the DODAG roots that primarily serve in the Step 5: On the contrary, the node will neither update the rank nor
network as data collection points and have the potential to construct a send any DIO message if its new computed rank is higher than its old
DODAG. They are also commonly referred to in the literature as gateway one, but it will keep the previous preferred parent as the current
nodes that connect the LLN to the external world (i.e., the Internet or preferred parent.
other networks) where they can receive requests or to where the gath­ Step 6: This process will be repeated by all neighboring nodes, as it
ered data will be managed. Router nodes are mainly located between the does in step 2, until all RPL nodes have joined the DODAG and
terminated when the leaf node is reached.

13
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Step 8: Another case occurs in the upward route when a new node
intends to join an already defined DODAG. In this case, the node asks
for the DODAG information from its neighbors by sending a DIS
message to request DIO messages that involve all the information in
the DODAG. When the new node receives the DIO messages,
depending on the OF, it will choose one or more parent nodes and
then select its preferred parent that has a lower rank.

As mentioned earlier, RPL nodes use a trickle timer to send DIOs

bearing in mind that this timer is reset when an inconsistency is detec­
ted. The trickle timer, which is crucial for maintaining network stability,
controls the rate of DIO messages. It is worth emphasizing that if this
rate goes down, the stability of the network will be better. Nevertheless,
when the RPL network is inconsistent, the intervals of DIO messages are
reduced to a minimum level (Mohammadinejad and Mohammadhoseini,
2019). The flowchart shown in Fig. 6 summarizes this process.

• Downward Route

The RPL uses DAO messages to construct and maintain downward

routes. Downward routes support only two types of traffic flows: P2MP,
which allows messages to flow from the DODAG roots towards the leaf
nodes, and P2P, which allows messages to flow from nodes towards the
DODAG root (or a common ancestor) via an upward route, and subse­
quently away from the DODAG root to a final destination via a down­
ward route. In fact, RPL supports two modes in which a RPL instance
may choose to maintain downward routes (Tsvetkov, 2011):

1. Storing Mode: In this mode, which is fully stateful, nodes keep

downward routing tables for their own sub-DODAG. On each hop in a
downward route, a node checks its routing table to determine the
next hop (see Fig. 7(a)). In storing mode, the RPL uses the IPv6
destination address to route messages downward. Furthermore, the
DAO is sent as a unicast message from the child node to the selected
parent(s) to enable multipath (see Fig. 7(a)). Each parent receiving a
DAO message stores the prefix and the child’s address in its own
routing table as the next hop prefix. The parent may respond to the
Fig. 6. DIO message communication and processing during the DODAG
construction.
DAO by sending a DAO-ACK to the sender of the DAO. It may also
delay sending a DAO message until it is aggregated with other DAOs
received from other nodes, and then sends it as a new DAO message
Step 7: Note that the leaf nodes can only join the DODAG but should
to its preferred parent. It is worth noting that a typical storing node
not transmit DIO messages except in some special cases, such as
will employ a single Transit Information option without a parent
detecting an inconsistency when forwarding a packet, where in this
field, and will send a DAO message so formed to one or more parents,
case, they have to send DIO messages to repair the inconsistency
with further Path Control modifications. When a node in storing
(Tsvetkov, 2011).

Fig. 7. Storing mode (a) and non-storing mode (b) for the RPL downward routing.

14
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

mode receives a DAO message with updated information (such as loops as they occur. In reality, this means that RPL guarantees neither
there is a new Path Sequence in the Transport Information option), it the loop-free path selection nor the narrow delay convergence times, but
must create a new DAO and unicast it only to each one of its DAO it can detect and repair the loop once it is used. This loop detection is
parents. Additionally, when a node removes a specific node from its used by RPL to assure that packets are making forward progress in the
own DAO parent set, it must send a special DAO message (called “No- DODAG Version and to trigger repairs when necessary.
Path DAO”) to that removed node to invalidate the current path.
Hence, this node attempts to create a new routing path by trans­ 1. Loop avoidance: A loop in RPL may happen when a node detaches
mitting a new DAO message through its new parent. The No-Path from a DODAG for various reasons and then attaches to a device in its
DAO message has a Path Lifetime of zero (0 × 00) in the Transit previous sub-DODAG (Tsvetkov, 2011). This may occur specifically
Information option and actually means loss of reachability. More when DIO messages are lost. Various rules have been identified to
precisely, there are three cases where a node removes its parent and avoid loops (Gaddour and Koubâa, 2012). These rules are based on
may use No-Path DAO messages. The first case is when the current the rank property of the RPL nodes. The max depth rule dictates: that
parent becomes unavailable due to a parent node failure. The second a node cannot choose a parent node whose rank is higher (i.e.,
case is when a node finds another parent node with potentially better deeper) than the node’s lowest rank + DAGMAXRankIncrease. As a
metrics than the current one. The last case is when the node turns matter of fact, this rule does not stop loops from happening, but it
into a new parent that it “thinks” has a superior metric but actually does avoid count-to-infinity when creating a loop. RPL as well pre­
does not (Jadhav et al., 2017). vents nodes seeking alternate parents to raise their ranks by choosing
2. Non-Storing Mode: In this mode, which is fully source routed, nodes deeper parents, as this is very likely to lead to loops in the LLNs.
do not keep downward routing tables for their own sub-DODAG. Another technique to avoid loops is for a node to poison its sub-DAG
Instead, downward packets are routed using source routes that are by declaring a rank of INFINITE RANK instead of using
populated by the DODAG root. In non-storing mode, the RPL uses the DAGMAXRankIncrease. On the other side, a DAO loop may happen
IP source routing to route messages downward. Moreover, the DAO when the parent node has a route installed when receiving and
is sent as a unicast towards the DODAG root (see Fig. 7(b)). When a processing DAO messages from a child node, but the child node has
node in non-storing mode receives a DAO message, it must spread the then cleaned up the related DAO state. This loop occurs when a
updated downward route information upwards. The downward No-Path is lost and continues until all states are cleaned up. Indeed,
route information might be aggregated with other DAO messages RPL involves an optional mechanism for acknowledging DAO mes­
before it is propagated upwards, which may necessitate propagation sages, which can help reduce the impact of losing a single DAO
delays. In addition, when a node removes a specific node from its message. RPL involves loop-detection mechanisms that reduce the
own DAO parent set, it may create a new DAO message with the effect of DAO loops and stimulate their repair (Tsvetkov, 2011).
updated Transit Information option. It must be noted that a typical 2. Loop detection: Because loops are difficult to avoid, RPL defines
non-storing node will employ various Transit Information options, mechanisms for detecting loops and resolving them when they occur.
and will send a DAO message so formed directly to the root. These mechanisms are based on the concept of data-path validation,
which is to carry routing information in the data packet to guarantee
3.2. DODAG detection and repair phase that the packet is moving in the forward direction and that no loop is
experienced. RPL loop detection uses this routing information using
In this subsection, we explain how the RPL addresses the DODAG the RPL Option in the IPv6 Hop-by-Hop Option header. In fact, the
failures. RPL Option holds RPL Packet Information and specifies the details of
how RPL packet information is transferred within the RPL Option.
• Maintenance of Routing Adjacency RPL uses the data packet for on-demand loop detection, thus the RPL
Packet Information, which is transmitted with the data packet, in­
In RPL, the concept of local confidence is used to evaluate when a cludes the transmitter’s rank. The potential loop is indicated by the
connection with another node (i.e., router or leaf) is legitimate and inconsistency between the routing decision of a packet (upward or
utilizable to construct a router adjacency, as well as to notify that an downward) and the rank relationship between any two nodes. When
extant RPL connection should be decommissioned. Notably, whether it such a packet is received, the node starts a local repair operation. For
is up the default routes or down along the DODAG, selecting successors example, suppose a node receives a packet marked as moving in the
induces routing adjacencies that must be repaired. Unlike other routing upward direction. If that packet registers that the rank of the sender
protocols, RPL does not use keep-alive events, which means that routing is lower than that of the receiving node, then the receiving node is
adjacency failures are not verified. This is due to the costly keep-alive able to infer that the packet did not forward in the upward direction,
operation in terms of bandwidth and energy (since a device that oper­ and also that the DODAG is inconsistent. Consequently, the receiving
ates on low-battery power cannot provide periodic keep-alive trans­ node performs a local repair (Tsvetkov, 2011).
missions). Thus, RPL needs an external procedure to notice the absence • Repairing Process
of a previously found neighbor. This external procedure must be
responsive to traffic to concentrate on the most used links and minimize Repair mechanisms are of dominant importance for any routing
the overhead of maintaining routing adjacency. RPL implements two protocol to adjust the network topology and dynamically update routing
reactive mechanisms, including Layer 2 triggers (obtained from associ­ decisions when link and node failures occur. To this end, RPL introduces
ation states or any Layer 2 related event), Layer 2 acknowledgments two complementary repair mechanisms, namely, local and global repair.
(Teraoka et al., 2008), and the neighbor unreachability detection The local repair operation is triggered when a network inconsistency is
mechanism (Narten et al., 2007). detected, such as a link failure, a loop, or a change in a DODAGID. It
actually consists of resolving detected inconsistencies by quickly
• Loop Avoidance and Detection discovering a backup path without attempting to repair the entire
DODAG. For example, when a node loses connection with its parent, it
Loops are a common undesirable problem in distance vector routing chooses a new parent. The local repair mechanism may not produce
protocols, including RPL. The RPL protocol avoids the formation of optimal solutions due to many inconsistencies. In this case, a global
loops by using rank-based data-path validation mechanisms to detect repair operation is needed as it is triggered by the DODAG root, which in

15
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

turn increments the DODAG Version Number, and thus initiates a new 3.3. Data forwarding phase
DODAG Version. Indeed, the global repair operation causes an essential
reconstruction of the topology, but this comes at the cost of decreased The RPL protocol supports both P2MP as well as MP2P communi­
network performance since the graph needs to be recalculated, which cations, whereby upward and downward routes are created and main­
increases network control traffic. Since the DODAG will be rebuilt from tained proactively. By creating these routes, RPL also supports P2P
scratch, the nodes’ positions in the new DODAG Version may differ from communication. When using P2P communication in storing mode,
those in the old DODAG Version. Therefore, the rank of the nodes may communication between any two nodes is attained by sending packets
also vary. upward till a common ancestor is reached, and then downwards from
that common ancestor towards the destination node. In non-storing

Fig. 8. An example of a DODAG construction using a path length metric.

16
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

mode, the packet must travel through the sink node since the interme­ Step 2: Nodes B and E receive the first DIO message from node A
diate nodes do not keep routing information. (which has the lowest rank) and then process the DIO accordingly.
Therefore, nodes B and E select node A as their preferred DODAG
3.4. Novel examples of DODAG construction parent and add node A to their parent list. Then nodes B and E
compute their new rank based on the path length, and update their
In this subsection, we provide illustrative examples of the DODAG DIO message with their new rank value of 2.
construction process using five routing metrics, namely: path length, Step 3: Node B broadcasts its own DIO message to all its neighboring
residual energy, end-to-end delay, ETX, and RSSI. Algorithm 1 clarifies nodes, i.e. A, C, F, G, and K. Since node A has a lower rank than that
how the DODAG is constructed in our examples. It is noteworthy to of node B, it does not process the received DIO message, but instead
mention that, in this algorithm, the node, which receives the DIO mes­ discards this DIO message. Nodes C, F, G, and K receive the first DIO
sage, has two options: either processes the message and selects the message from node B and choose it as their preferred parent, and
preferred parent that satisfies the OF, or discards the DIO message since then broadcast their updated DIO message with their new rank value,
it comes from the parent that has a higher rank since RPL employs the which is 3.
notion of rank to avoid loops. Depending on that, the routes are estab­ Step 4: Node E also broadcasts a DIO message to neighboring nodes
lished between each node and the DODAG root. A, F, I, and J. Node A discards the received DIO message because it
has a lower rank than node E. Since node F receives the second DIO

Algorithm 1. Pseudocode for DODAG construction in RPL.

message from node E, it processes the DIO message by computing its

In our first example (Fig. 8), we show how the DODAG is constructed new rank and then comparing it with its previous one. Node F finds
based on the path length. Each node will choose the preferred parent that its distance on the path A → E → F, DA→E→F , is 2 m, which is less
with the lowest path length. The steps below describe constructing a than its distance to its preferred parent node (i.e., node B) on the path
DODAG using this routing metric. A → B → F, DA→B→F , which is 5 m. Therefore, node F changes its
preferred parent to be node E, updates its rank and the DMC in a DIO
Step 1: The DODAG root (i.e., node A) starts sending link-local message, and then broadcasts the new DIO. On the other hand, nodes
multicast DIO messages to its neighboring nodes. One of the nodes I and J receive the first DIO from node E, so they choose it as their
can further decide to send a DIS message. In this case, Node A will preferred parent and broadcast their DIO message with their
immediately send a DIO in response to a DIS message. computed rank (rank = 3).

17
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Fig. 9. An example of a DODAG construction using a residual energy metric.

Step 5: Node C broadcasts a DIO message to nodes B, D, and G. Node message because the DA→E→J→I > DA→E→I as well as the
B discards the DIO because it has a lower rank than node C. Node G DA→E→J→K > DA→E→F→K .
receives the second DIO from node C, where the DA→B→C→G is 6 m, Step 9: Node D broadcasts a DIO message to nodes C and H. Only
which is less than the previously recorded path length, DA→B→G = 7 node H chooses node D as its preferred parent.
m. Therefore, node G chooses node C as its preferred parent rather Step 10: Node G also broadcasts a DIO to nodes B, C, H, and K. Nodes
than node B and updates its rank from 3 to 4. B and C ignore the received DIOs because they have lower ranks.
Step 6: Node F broadcasts a DIO message to neighboring nodes B, E, Similarly, node K ignores the message as its DA→E→F→K when node F is
and K. Nodes B and E discard the DIO message because their rank is its preferred parent is 3 m, which is less than its DA→B→C→G→K , which
lower than that of node F. Since the preferred parent for node K is equals 7 m, when node G becomes its preferred parent. Conse­
node B, node K compares its DA→B→G→K , which is 8 m, with the quently, node G becomes the parent of node H.
DA→E→F→K , which is 3 m, to select the preferred parent among its Step 11: Node K broadcasts its DIO message to surrounding nodes B,
neighbors. Node K finds that the DA→B→G→K > DA→E→F→K , accord­ F, G, J, and L. Nodes B, F, and J ignore the DIO message because node
ingly, node K changes its preferred parent to node F and its new rank K has a higher rank than them. Node G finds that the
becomes 4. DA→B→C→G > DA→E→F→K→G , so it changes its preferred parent to node
Step 7: Node I broadcasts a DIO to nodes E and J, who in turn discard K instead of node C, and its rank becomes 5. Node L also chooses
this message. The reason again is that the rank of node E is less than node K as its preferred parent.
the rank of node I. Also, the DA→E→J = 2 m is smaller than the Step 12: Because node G has updated its preferred parent and its
DA→E→I→J = 3 m, which means that it is better to keep the previous rank, it sends the updated information in a DIO message to neigh­
preferred parent of node J (i.e., node E) as the current preferred boring nodes B, C, H, and K. Based on this received DIO message,
parent rather than replace it with node I. node H updates its rank to 6, while nodes B, C, and K do not process
Step 8: Node J broadcasts a DIO message to the neighboring nodes E, this DIO.
I, and K. Node E discards the DIO because its rank = 2, which is less Step 13: Node L broadcasts a DIO message to nodes K, H, and M.
than the rank of node J. Nodes I and K also discard the received Node K ignores the received message because its rank is lower than
that of node L. Since DA→E→F→K→L→H = 7 m and DA→E→F→K→G→H =

Fig. 10. An example of a DODAG construction using an end-to-end

delay metric. Fig. 11. An example of a DODAG construction using an ETX routing metric.

18
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

5 m, node H keeps node G as its preferred parent and discards the preferred parent. In addition, node F chooses its preferred parent to be
DIO message from node L. In this case, only node M chooses node L as node C because it presents the lowest ETXA → C → F , which is 5.25. Nodes
its preferred parent. G and H find that node F is the best choice to be their preferred parent
Step 14: Node H broadcasts its DIO message to its neighboring because it guarantees the lowest ETX value. More precisely,
nodes, i.e., D, G, L, and M. Nodes D, G, and L have lower ranks than ETXA → C → F→H = 8.75 and ETXA → C → F→G = 6.25. Finally, node I has
node H, thus they ignore the DIO message. Node H becomes the the option to choose either node G or node H as the preferred next hop.
preferred parent of node M. Since ETXA → C → F→G→I < ETXA → C → F→H→I , node I chooses node G as
Step 15: Demonstrates the final DODAG after it is constructed. the preferred parent node.
Achieving reliability is crucial in LLNs due to the varying environ­
The second example (Fig. 9) shows the construction of a DODAG mental conditions that influence the IoT networks, such as interference
based on the residual energy. In this example, we estimate the node’s and rain. The need for assessing the wireless medium and links’ quality
residual energy on a scale from 255 mJ (full) to 0 mJ (empty). is believed to be in demand. Reliability is evaluated through link esti­
Furthermore, we assume that the DODAG root has the maximum energy mators such as the RSSI metric. RSSI is a hardware-based link estimator
value. Each node will choose its preferred parent with the highest re­ and PHY (layer 1) metric. The “PACKETBUF_ATTR_RSSI” variable from
sidual energy. When considering the topology shown in Fig. 9(a), we can the packet buffer file is used to read the RSSI in Contiki. RSSI’s range is
observe that node F can choose either node B or D as its preferred parent. between [− 100, − 50] dBm. Higher values of RSSI denote more reliable
Since the residual energy of node B (ERnode B = 220 mJ) is greater, node links and vice versa. To obtain the correct value of RSSI according to the
F will choose it as the preferred parent node. Likewise, node H chooses datasheet of motes in Contiki, we add an offset of (− 45), which is the
node E instead of node F as its preferred parent because its residual front-end gain, using the following formula True RSSI Value =
energy (ERnode E = 218 mJ) is greater than the residual energy of node F Last received packet RSSI + ( − 45), which shows how we find the cor­
(ERnode F = 210 mJ). The other nodes choose their preferred parent in rect RSSI based on receiving one packet. However, the average RSSI is
the same way. calculated after receiving a bundle of packets. In Fig. 12, we present the
Consider the topology depicted in Fig. 9(b), in which node B last example that shows the construction of a DODAG based on the RSSI
consumed approximately 25 mJ of its energy while the other nodes metric. The node chooses the path with the highest average RSSI value
consumed only 5 mJ. We can see that node F changes its preferred parent (best LQ), as the node gets the RSSI value from a DIO message. We can
from node B to node D, which has higher residual energy. Node E also see in Fig. 12 that node G can transmit its data message to either node E
detects that the residual energy of node B (ERnode B = 195 mJ) has or node F. Node G computes the average RSSI value of all nodes on the
become less than that of node F (ERnode F = 205 mJ). Therefore, node E path of node E and for all nodes on the path of node F towards the
switches its previous preferred parent to node F. DODAG root. It finds that the average RSSI on the path A → C → E→G,
The example in Fig. 10 shows the construction of a DODAG based on μRSSIA → C → E→G , is − 70 dBm, and the average RSSI on the path
the end-to-end delay metric. In this case, the DODAG root has a 0 ms A → B → D→F→G, μRSSIA → B → D→F→G , is − 68 dBm. Accordingly, node G
value. On the other hand, other nodes have varying delays. Each node chooses node F as the preferred next hop (i.e., the preferred parent).
will choose the preferred parent with the shortest end-to-end delay. As
an instance, nodes E and F can choose either node B or node D as their
preferred parent. Since node B has a less end-to-end delay (EED = 3.5. Novel examples of DODAG repair mechanism
0.2 ms) than node D (EED = 0.4 ms), node B will be the best choice for
nodes E and F to become their preferred parent. Furthermore, both Here, we provide an illustrative example of the local repair mecha­
nodes G and H will choose node E as the preferred parent because it nism implemented by RPL when a link failure occurs within the same
∑
provides the lowest delay on the path A → B → E (i.e., DODAG. In particular, we consider two cases of the network topology
EEDA → B → E = 0.5 ms) compared with the path A → B → F, which has shown in Fig. 13 to demonstrate how this mechanism is triggered in
∑ ∑
EEDA → B → F = 0.6 ms. Finally, EEDA → B → E→H = 0.75 ms, response to link failure. In case 1 (Fig. 13(a)-13(d)), we assume that the
∑
which is smaller than EEDA → B → E→G = 0.825 ms. Thus, node I picks link between node D and node F is broken and there is no alternative
node H as the preferred parent. path from node F towards the DODAG root. More precisely, when node F
Let us consider another example described in Fig. 11, where a detects losing its connection to neighboring node D when forwarding a
DODAG is constructed based on the ETX routing metric. Each node picks data packet, it removes node D from its parent list and starts the local
the preferred parent that has a path with a minimum ETX value. As repair process in order to find an alternate parent/path. Since node F has
detailed in this figure, node E may choose either node B or node C as the lost all its parents and its DODAG parent set is empty, it becomes the
preferred next hop. Since the cumulative ETX on the path A → C → E, DODAG root of its own floating DODAG. Thus, node F directly an­
ETXA → C → E , is 7.25 which is less than the cumulative ETX on the path nounces this new situation in the DIO message as seen in Fig. 13(b).
A → B → E, ETXA → B → E , which is 9, node E picks node C as the After node G receives this DIO message as a link failure notification, it
detects that node F has detached from the DODAG and can no longer be
considered its parent. As a consequence, node G changes its preferred
parent to node C and removes node F from its parent list. When node H
also receives the DIO message from node F, nodes H, I, and J become the
children of node F in its sub-DODAG (see Fig. 13(c)). At this point, node
F acts as a floating DODAG root with a rank of 1, while nodes H, I, and J
update their ranks to 2, 3, and 3, respectively. Finally, in Fig. 13(d), node
G broadcasts its own DIO message to notify other nodes of its new
preferred parent (i.e., node C). Upon receiving this DIO message, node F
decides to join the grounded DODAG and becomes the child node of
node G with a new rank of 6. Nodes H, I, and J adjust their rank
accordingly to be 7, 8, and 8, in turn.
On the other hand, in case 2 (Fig. 13(e)-13(h)), we assume that the
link between node D and node F is broken and there is an alternate path
between node F and the DODAG root, which is A → E → H→ F. In this
Fig. 12. An example of a DODAG construction using the RSSI routing metric. case, node H knows that the connection to a grounded DODAG is

19
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Fig. 13. An example of a DODAG local repair mechanism when considering two cases in which the link failure occurs between node D and node F. Case one, (a) to
(d), there is no alternate path from node F to node A. Case two, (e) to (h), there is an alternate path from node F to node A, which is the H-E path.

possible through the node E. Based on this information, node H unicasts

a DIS message to node E to rejoin the DODAG, which in turn responds
with a DIO message (see Fig. 13(e)). Consequently, node E becomes the
parent of node H, which in turn updates its rank to 3 and broadcasts the
updated DIO message to its neighboring nodes E, F, I, and J, as shown in
Fig. 13(f). Likewise, nodes I and J update their rank with the same value,
which is 4, and then they join the grounded DODAG through Node H.
Once node F receives the updated DIO from node H (see Fig. 13(g)), it
rejoins the DODAG through node H. Finally, in Fig. 13(h), node F be­
comes the child node of node H, and its new rank is 4.

4. RPL applications

This section defines RPL applications and their requirements, to

determine appropriate routing metrics and constraints, which are
translated into an OF.
Most of the RPL applications are inherited from the WSNs and IoT
applications, as RPL has mainly been designed for LLNs employed in IoT.
As a result, RPL routing solutions may be used in a broad range of ap­
plications, such as home automation (Brandt et al.), building automation
(Martocci et al., 2010), urban networks (Dohler et al.), and industrial
automation (Pister et al., 2009). In some cases, the adaptability of the
RPL must be taken into consideration to enhance network performance.
Fig. 14 shows the top seven popular RPL applications at the moment.

Fig. 14. The most prominent applications of the RPL.

20
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

4.1. Smart grid (SG) understand the effect of light earthquakes on all city buildings
(Zanella et al., 2014).
SGs have been around for more than a decade and are essential to the • Waste management: Waste management considered as a funda­
digital transformation of the electrical industry. To be more specific, SG mental concern in each modern city, due to the cost of service and
is an electrical network that uses digital communications technology to the problem of garbage storage in landfills. However, deeper infil­
allow for a two-way flow of electricity and data, as well as the detection, tration of ICT solutions for this area may lead to significant savings
reaction, and prevention of changes in usage and other problems. SGs and economic and environmental benefits. For example, the adop­
are self-healing and give subscribers a say in how the system works. tion of smart waste containers, which notice the load level and allow
Thus, automatic control of energy consumption and feedback responses optimization of the routing of the collection trucks, can decrease
have gained great interest in academia and industry. For this purpose, waste collection costs while also improving recycling quality. To
the SG may balance the energy distribution depending on the amount of achieve a smart waste management service like this, the RPL-based
power used. The SG mainly consists of a power system, AMI, and a IoT must connect the end devices (such as smart waste containers)
network called the SG communication network, which is separated into to a control center, where the optimization program processes the
various network zones, including industrial, homes, and neighbors data and decides the best management of the collection truck fleet
(Feroz and Ikram, 2020). Regarding the SG, within the neighbor area (Zanella et al., 2014).
network that contains devices at sites and utility monitors, the authors in • Air quality and noise monitoring: In today’s world, air and sound
(Liu et al., 2017) have incorporated parents’ switching methods into the pollution are becoming an increasing threat. With a focus on the
RPL design, which permit nodes to switch preemptively between health of living organisms, it is critical to detect and maintain air
different parents due to the necessity of having a body of trials to receive quality within acceptable levels. Air monitoring techniques are used
intended packets. Many countries, notably Australia, South Korea, to calculate concentrations of hazardous gases. Then, the data is
Japan, and China, have funded extensive investments in the next gen­ analyzed statistically to allow the authorities to take action to pre­
eration of grid technology. By 2020, the European Union will deploy vent pollution. As a result, air quality is created in such a way that
smart meters to more than 80% of its customers. Other countries are people can choose the healthiest, cleanest, and safest area for open-
directly exploiting the SG or developing the renewable energy sector air activities while still being connected to their preferred exercise
indirectly, which will eventually require more sophisticated SG tech­ applications through the RPL-based IoT ecosystem (Maisonneuve
nology (Borlase, 2013). There is no doubt that RPL plays a prominent et al., 2009).
role in SG deployment, and it has the potential to be the standard routing • Traffic congestion: In the not-too-distant future (2030), two-thirds
protocol in AMI applications. AMI (Aijaz et al., 2015) investigated the of the world’s population will live in urban areas, putting enormous
rank-calculation and failure-handling methodologies (Abdel Hakeem pressure on the city’s infrastructure. As a result, urbanization has
et al., 2019). become a major concern (Meredith, 2018). Accordingly, ICTs are
deployed to provide and make smart city services more accessible to
individuals, including traffic control, to improve urban resource ef­
4.2. Smart cities ficiency while reducing environmental effects and contributing to
economic prosperity. In light of this, a number of cities have sought
Smart cities have captured the imagination of academia, industry, to reduce traffic congestion by implementing smart services such as
and governments around the world in recent years, resulting in a large smart traffic signs and smart parking systems. Intelligent trans­
number of projects and research. Without the IoT, the design of a smart portation systems can also make travel safer, more convenient, and
city will be unachievable. To that end, the use of IoT in urban areas is more secure. This can be achieved by using IoT sensors and routing
becoming increasingly ubiquitous. This is because many national gov­ protocols that provide effective routes for the sake of mitigating
ernments are moving towards the adoption of ICT in administering traffic congestion, ensuring free-parking zones for people, and pre­
public affairs, therefore achieving the so-called “smart city” concept venting accidents through rapid response to such events (Sethi and
(Zanella et al., 2014). The aim of a smart city is to maximize the use of Sarangi, 2017). Another issue that has drawn attention to trans­
public resources, improve the QoS provided, and reduce the operational portation in smart cities is the widespread use of self-driving vehi­
costs of many traditional public services, including transportation, cles. To emphasize the points of these challenges, various factors
lighting, parking, surveillance, and garbage collection. Moreover, the must be considered when implementing RPL in such applications,
information collected by IoT devices in a smart city can also be exploited including real-time information, security, mobility support, and
to enhance people’s awareness about the situation of their city as well as reliability. However, security is a critical factor as any attack may
stimulate the origination of new services. Actually, these services can threaten human lives. Despite the fact that RPL is commonly used to
use RPL with different OFs that meet the application requirements. Here transmit data in static on-road nodes, few research papers have
are some of the services that smart cities may provide (Khalifeh et al., addressed its application in the vehicular network with mobile
2021): nodes. Moreover, integrating the pair, IEEE 802.15.4 and RPL, to be
used in vehicular ad-hoc networks is still arguable (Kharrufa et al.,
• Structural health of buildings: The appropriate maintenance of the 2019).
city’s historic buildings requires constant monitoring of the real
conditions of each building as well as the identification of areas most 4.3. Smart home (SH)
vulnerable to the influence of external agents. The RPL-based IoT
sensors located in buildings obtain building structural integrity The SH is an important part of the IoT. It provides effective service to
measurements (Loh and Lynch, 2006). Various types of such sensors clients by establishing communication with various digital devices
are used inside buildings, i.e., vibration and deformation sensors that within them. This technology connects people and things anytime and
monitor building stress, temperature and humidity sensors that anywhere, making life easier, more comfortable, secure, and practical
provide a complete picture of environmental conditions, and for individuals. Smart devices are installed and controlled by a central
weather sensors in nearby areas that monitor pollution levels. Data station in a SH. This station is controlled by the user through mobile
collected by these sensors helps decrease the need for costly periodic phone applications. Heating, cooling, and air conditioning systems, as
structural examination by human hands and permit targeted and well as security systems and lights, can be found in SHs. Sensors and
proactive maintenance as well as recovery actions. Also, the vibra­ switches (known as gateways) that communicate with a central station
tion and seismic readings may be combined to better investigate and or are operated by IoT technology are used in these systems (Alaa et al.,

21
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

2017). On the other hand, SH conserves energy by improving domestic noticeable RPL improvement in terms of energy balancing and fast local
energy management techniques (Hargreaves et al., 2017). The RPL repair. It also improves the packet delivery ratio, minimizes packet loss
protocol can be applied to improve its performance as it helps to reduce ratio, and reduces end-to-end delay in the DODAG root. Additionally,
energy consumption in IoT network. the proposed method has significant improvements in multi-path rout­
ing compared with the standard RPL HC. The authors confirm that the
4.4. Military RPL satisfies the criteria in greenhouse circumstances using this
methodology.
The military is a sensitive field that makes implementing and
developing applications a critical and challenging task for any tech­ 4.7. Smart manufacturing system (SMS)
nology. It is often challenging to access the nodes after deployment. In
such applications, reliability, mobility support, efficient use of energy, The ability of manufacturers to compete in terms of quality, cost,
and security are considered key challenges. Without these challenges, delivery, and flexibility determines the outcome of competition. By
both active and passive monitoring can become very restricted, and using advanced technologies and combining communication and infor­
misleading data can be produced, causing severe harm. Replacing the mation technologies with intelligent software applications, SMS im­
nodes’ batteries is an almost impossible task, especially in war zones and proves manufacturers’ capabilities. These technologies are employed for
dangerous places. Thus, energy consumption becomes a major concern. two purposes, namely: (i) maximize the exploitation of materials, en­
Military applications rely on sensor nodes, whose aim is to provide more ergy, and labor; and (ii) respond quickly to changes in market demand
safety for both soldiers and personnel by offering monitoring informa­ and supply chains. SMS is unique in that it regulates a vision of the next
tion, emergency navigation, disaster deterrence, and robotic interfer­ manufacturing generation with improved abilities (Lu et al., 2016). SMS
ence. Using wearable sensors, RPL can also be used to detect mines (Niu systems are tolerant of new scenarios because they make intelligent
et al., 2007) and determine the physical condition of soldiers to find decisions based on real-time data. Furthermore, anticipating and
degrees of fatigue (Javaid et al., 2013). avoiding failure is critical. To that end, SMS may use remaining energy
and end-to-end delay as routing metrics for RPL. Moreover, RPL may use
artificial intelligence algorithms for SMS.
4.5. Medical and healthcare

The evolution of the IoT and ubiquitous sensor nodes enables the 5. Simulation platforms supporting RPL
implementation of medical applications that take care of the in­
dividual’s health. Thus, remote health monitoring is one of the most This section provides the readers with the most commonly used RPL
advanced technologies used by medical experts to monitor a patient’s simulation platforms and their status in a comprehensive table. In
vital health indicators on a regular basis. Many factors influence health addition, we review the most popular one (COOJA) in-depth. In the
applications, such as security, end-to-end delay, and mobility, since a course of RPL’s simulators, we have addressed the most up-to-date in­
patient uses wearable sensors to track his movement. Therefore, these formation about the status of RPL over NS-3.
factors must be considered to effectively deliver the collected data from Until now, there have been several software platforms that support
the wearable sensors. More to the point, the more efficient the appli­ the implementation of the RPL, such as NS-2, OMNeT++, JSim, COOJA,
cation, the safer the lives of patients. Thus, RPL is the most suitable etc. However, some of these platforms are not suitable for large-scale
routing protocol for these purposes. By way of illustration, the authors in networks (Patil and Hadalgi, 2012) (Khan et al., 2012). Table 6 (see
(Gara et al., 2015) investigated an application for dynamic and hybrid next page) summarizes a comparison of the most popular simulation
network topology and implemented a modified version of RPL. The platforms used to model the RPL protocol. COOJA (Sobhan babu et al.,
performance of this modified version is enhanced compared to the 2017), in which the RPL is fully implemented, is a cross-layer simulator
original version of RPL, particularly for the mobile node, which is for the Contiki OS that is specifically designed for devices with limited
essential in these applications. They suggested implementing mobile resources, such as IoT devices and WSN sensor nodes. In particular, it
nodes as leaf nodes and only transmitting DIS messages to join the simulates different network layers and also supports the hardware
DODAG and then becoming a member of the parent node list without simulation of IoT sensor nodes (Jadhao and Solapure, 2017). This
disseminating DIO messages. simulator creates graphs for a variety of network statistics as well as a
timeline for the transmission of each node (Mehmood, 2017). Contiki
(Gaur and Tahiliani, 2015) is an open source, lightweight, and highly
4.6. Agriculture greenhouse
portable OS written in the C programming language and specifically
built to run on hardware devices that are severely restricted in power,
The greenhouse environment can be monitored using the multi-path
memory, and communication bandwidth. Contiki has three network
RPL protocol as suggested in (Thu et al., 2015). By utilizing HC and
stacks: (i) the uIP TCP/IP stack for IPv4 networking, (ii) the uIPv6 stack
residual energy as routing metrics in the RPL protocol, reasonable re­
for IPv6 networking, and (iii) the Rime stack for LLNs (Solapure and
sults are granted in greenhouse scenarios. Precisely, the suggested
Kenchannavar, 2020). The Contiki NETSTACK is mainly divided into
scheme, when employed in realistic greenhouse networks, records a
four layers, as shown in Table 5: (i) the Network layer (NET­
STACK_NETWORK), which contains the Application, Transport,
Table 5 Network, and Adaptation layers, (ii) the MAC layer (NETSTACK_MAC),
Contiki network stack. (iii) the RDC layer (NETSTACK_RDC), and (iv) the Radio layer (NET­
Layer Protocol Files in Contiki OS STACK_RADIO). All NETSTACK layers run on the packet buffer “pack­
Network Application IETF CoAP/REST websocket.c, http-socket.c, etbuf”, which is large enough to maintain one radio packet that, of
Engine coap.c course, represents the current packet. In addition, queue buffers
Transport UDP udp-socket.c, tcp-socket.c “queuebuf” are used to store received packets in queues for later use
Network, IPv6/RPL uip6.c, rpl.c
(Mugdhe, 2016).
Routing
Adaption 6LoWPAN sicslowpan.c At the network and routing layer, the Contiki automatically creates a
MAC CSMA/link-layer nullrdc.c, contikimac.c wireless IPv6 network utilizing the RPL routing protocol. It is worth
bursts noting that the network layer is separated into two sub-layers: the upper
RDC ContikiMAC csma.c IPv6 layer and the bottom adaptation layer. The adaptation layer pro­
Radio IEEE 802.15.4 cc2420.c
vides IPv6 and UDP header fragmentation and compression techniques.

22
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 6
Comparison of RPL simulators.
Simulator Developed Support for RPL Support for Supported Programming Timing Wireless channel Ref.
year Large-scale platforms languages
simulations

NS-2 1989 Object-oriented model, - Adopt less General C++ and OTcl Discrete Free space, two-ray Issariyakul
which is open to than 100 nodes simulator. event ground refection, and and Hossain
implementing new - Complicated. Compatible with shadowing. (2012)
protocols. Linux, Ben Saad
Flexible for overall WSNs Mac, and et al. (2011)
simulation. Windows.
It was unable to simulate
some WSN issues, such as
energy consumption,
bandwidth, and energy
conservation.
OMNeT++ 1997 Offer different frameworks Free scale General C++ and NED Discrete Lognormal shadowing, Varga (2001)
to diffuse the network with simulator. simulator. language event experimentally Ben Saad
RPL. Compatible with measured. et al. (2011)
Flexible for overall WSNs Linux, Path loss map packet
simulation. Mac, and delivery rates map.
Provide models for mobility Windows. Temporal variation and
and energy consumption. unit disk
JSim 2000 Support for different routing Adopt around -Compatible with Java and tcl Discrete n/a Tyan et al.
protocols. 500 nodes. Linux, Mac, and script language event (2003)
Support only the IEEE Longer time Windows.
802.11 MAC protocol, execution.
which is not compatible Complex.
with RPL. implementation
Inactive since 2006.
COOJA 2002 Part of Contiki OS. Adopt around Compatible with Standard C Discrete Multipath ray tracing Voigt (2012)
Fully supporting RPL. 200–500 Linux, Mac, and event with support for Ben Saad
nodes. Windows. attenuating for et al. (2011)
Long execution obstacles.
time. Unit disk and directed
Relatively low graph.
efficiency.
TOSSIM 2002 TinyRPL provides all kinds Adopt TOSSIM is a C++ and Discrete Lognormal shadowing. Levis et al.
of traffic patterns in RPL thousands of specific simulator python event (2005)
(P2P, MP2P, and P2MP). nodes for TinyOS Levis and Lee
TinyRPL is not supported on applications to be (2003)
the TOSSIM simulator, executed on Ben Saad
which needs a micaz binary. MICA Motes. et al. (2011)
Thus, it is not a fully
supported RPL simulation.
NS-3 2008 Handling the design of Free scale General C++ and Discrete IEEE 802.11 and YANS. Carneiro
Internet protocols and simulator. simulator python scripts event Free space, two-ray, and (2010)
networks work models. - Compatible with shadowing. Chen et al.
The RPL model is ready for Linux, Mac, and Small-scale and path loss (2012)
use. Windows. model.
WSim/ n/a The RPL model performs Scale-free - Compatible with C language Discrete - Free space, two ray Online
WSNet DODAGs’ construction, rank simulator. Linux, Windows, event ground, and lognormal (2019)
calculation, and packet Solaris, and MAC. shadowing. Ben Saad
transmission. - Rayleigh fading, ITU et al. (2011)
The mobility model was indoor model, and
modified to permit root nakagami fading
nodes to move according to - File static and disk
different moving schemes. model.
The drawback with WSNet is
that it considers only the
upward routes, which
implies it uses the MP2P
traffic pattern only.
Open WSN 2012 - Handling a complete - Scale-free - Compatible with Python Discrete - IEEE802.15.4e Watteyne
network stack for WSN/IoT simulator. Linux, Windows, event standard Time et al. (2012)
- Fully supporting of RPL and OS X Synchronized Channel Sesli and
non-storing mode. platforms. Hopping (TSCH) and Hacıoğlu
IEEE 802.15.4 2006 (2017)
Watteyne
(2014)

The MAC layer is considered as the simplest one in the IoT/IP stack, MAC, and LPP. ContikiMAC is based on the principles of low-power
which is used to avoid collisions. There are two MAC protocols in listening but with better energy efficiency. Contiki’s X-MAC is based
Contiki OS called CSMA and nullmac. Furthermore, the RDC layer saves on the original X-MAC protocol, but has a number of enhancements and
power by allowing the node’s radio transceiver to be turned off most of extensions to save power and maintain stable network conditions.
the time. Contiki OS has three duty cycling mechanisms: ContikiMAC, X- Contiki’s LPP is based on the LPP protocol, but with improvements that

23
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

• rpl. c: It is the basic file that contains the implementation of RPL

functions for altering routing entries (Winter et al., 2012).
• rpl-timers. c: It makes the RPL route lifetime RFC compliant.
• rpl-icmp6. c: It contains the implementation of RPL control messages
(DIO, DAO, DIS, and DAO-ACK).

To this end, COOJA is a popular RPL simulator with a graphical user

interface, which makes it one of the most impressive tools out there.
Furthermore, users can use a simple platform to create their own pro­
grams. Nonetheless, COOJA has several drawbacks, including its
adherence to the Contiki OS architecture and the lack of optimization for
large-scale topologies (Chen et al., 2012). Another drawback is the lack
of supporting documentation (Thomson et al., 2016). Most notably,
there is a problem with the random position of nodes, based on our
extensive expertise. To put it another way, integrating it with Google
Maps to ensure realistic node placements is difficult. After conducting
Fig. 15. The implementation of RPL in Contiki OS. thorough research and testing on RPL simulators, we discovered that the
NS-3 simulator is the most straightforward in terms of Google Maps
reduce power consumption and enable broadcast data transmission integration.
(Mugdhe, 2016). Finally, the radio layer is the Contiki Netstack’s lowest In fact, the NS-3 is a discrete-event network simulator, whereas the
layer, where data arrives here in bytes or as a whole packet via interrupt purpose of the NS-3 project is to develop a preferred open-simulation
handlers. The input data is read into the buffer, packetbuf, and the environment for networking research. The first implementation of the
process is polled. The process will be sent to the special event and for­ RPL model was proposed in (Bartolozzi et al., 2012). Unfortunately, this
warded to the upper layers as a result of this polling process. project was not released as open source till now. Interestingly, the NS-3
Actually, Contiki provides an implementation of RPL, called Con­ supports the IPv6 model and 6LoWPAN adaption functions, which are
tikiRPL, which is demonstrated in Fig. 15. The RPL directory of Con­ the basis for the implementation of the RPL model implementation.
tikiRPL contains various files that implement the RPL functions. Some of Fig. 16 illustrates a simplified UML classes diagram of the RPL model in
these files include (Solapure et al., 2020): NS-3 (Bartolozzi et al., 2012) (Chen et al., 2013). The control messages’
header is inherited from the Icmpv6 Header and Icmpv6 Option Header
• rpl-mrhof. c: It describes how to build an MRHOF with ETX as the classes. RPL is integrated via the IPv6 Routing Protocol class. In addition,
routing metric. several essential classes were created, such as the Routing Table, the
• rpl-of0: It contains the implementation of0. Neighbors Set, the RPL Metrics, and the RPL OFs, along with other
• rpl-dag. c: It primarily includes the key functions for DODAG inherited classes from the recently mentioned classes.
construction. Based on the authors’ experience and to make the integration

Fig. 16. Simplified UML class diagram of RPL model in NS-3.

Fig. 17. The process of routes mobility model.

24
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

between the Google map and the NS-3 version 30 (ns-3.30) platform traffic pattern, while the multicast protocol is taken into consideration in
feasible, it is recommended to install the NS-3 package on Ubuntu 18.4 (Hui and Kelsey, 2016). More efforts are to be exerted on DAO sched­
(Online]. Available, 2018). Thereafter, it is recommended to install the uling to mitigate the congestion and buffer requirements. Additionally,
Routes Mobility Model, which mimics real mobility suggestions by when the size of the network increases, the storage issue will appear. The
enquiring about the directions between nodes from the Google Maps drawback of the storing mode is the high memory consumption, while
Directions API. The Google Maps information, used in tandem with the the high communication overhead is the downside of the non-storing
ns3::Way point Mobility Model, permits the simulator to use realistic mode (Zhao et al., 2017). The issue here is to come up with an effi­
mobility suggestions based on real-life locations and road networks cient solution by successfully, amalgamating both schemes to minimize
(Online]. Available, 2019). The Google Maps Directions service offers an the risk of memory usage and extend the exploitation of node capacity.
XML file that covers: (i) Leg, which holds a list of steps in a waypoint (A
to B, passing through C); (ii) Step, which has the points encoded in a 6.2. Energy consumption challenge
polyline; and (iii) Point, which is a latitude/longitude set (i.e.,
geographical location). Finally, iv) the direction data includes (but is not Limited energy is one of the most crucial issues facing the LLNs, as
limited to) a polyline for each step of the requested path. Consequently, the design of the IEEE 802.15.4 and the RPL have considered energy
the module decodes the polyline to create a list of geographical and then consumption and proposed approaches to reduce its use. Aside from
Cartesian coordinates. Intriguingly, Fig. 17 demonstrates the mecha­ scalability, network diversity and nodes heterogeneity increase energy
nism of the designed model. In particular, the user must first identify the consumption (Liu et al., 2017). As mentioned above, the energy con­
start and endpoints by sending them to the Choose Route function, which sumption issue of RPL is addressed by the trickle algorithm (Levis et al.,
in turn is requested by the Google Map API. After processing the request, 2011), which aims to lessen the number of needless control messages.
an XML file is sent back to NS-3, which manipulates the data and gen­ Despite this, the trickle has been shown to have its own shortcomings in
erates the points, steps, and legs. dealing with dynamic environments, leading to inefficient data trans­
Despite the easiness of addressing the aforementioned COOJA’s mission and significant energy loss due to packet delivery failures
limitation in NS-3, having an open-source of RPL on NS-3 is still (Cobarzan et al., 2014a) (Kharrufa et al., 2017).
ongoing. As a matter of fact, we are currently involved in a scientific The second problem is related to the different relative distances be­
group working on releasing a RPL open-source project. In particular, the tween a certain node and its root node in the network. This distance
code that is developed mostly works, but the design is not properly set. variation may result in considerable energy consumption among nodes,
In other words, the relationship between classes, the responsibility of and this can lead to congestion issues that influence network reliability.
each class, and the architecture of the code itself all still need revising. Recent researchers have attempted to take the energy depletion rate of
Among the RPL metrics, the HC is the only one that is currently working the nodes into routing metrics and generate predictions about the route
well. In addition, the testing issues for each class and procedure are still that will drain the least amount of energy. An alternative option is to
missing and should be perfectly performed to guarantee that everything introduce a backup node to replace dead nodes at the lowest network
is working as intended for most basic cases. cost. Nevertheless, bottleneck nodes (i.e., nodes that consume more
energy than other nodes) can be somewhat inevitable, so critical ques­
6. RPL challenges tions about how to effectively balance nodes’ energy are what should be
considered in the future.
Heretofore, we have discussed the RPL protocol from various aspects.
Although it is generally known as the standard routing protocol for IoT 6.3. Diversification of OFs challenge
networks, there are many challenges that RPL has faced since its
inception. In this section, we present and discuss some of these chal­ The OF in RPL is responsible for selecting and specifying the
lenges that need to be considered when designing new RPL-based pro­ preferred parent by calculating the rank and, thus, the best path to reach
tocols. Here are some of the key challenges: the root. However, suggesting an adequate OF in LLNs is quite chal­
lenging. Because of the nature of IoT applications, the performance of
6.1. Efficient routing support with limited memory challenge LLNs is limited in terms of the packet delivery ratio, end-to-end delay,
efficient energy, and network stability. Most of the available literature
Routing stability and traffic patterns such as P2P, MP2P, and P2MP has investigated the effective or influential elements in route formation,
with heterogeneous node capabilities, are frequent demands for LLNs including control message overhead, LQ estimations, and residual en­
applications (Levis et al., 2009). As mentioned earlier, the MP2P ergy. It has been demonstrated that the incorporation of the dominant
transmission traffic pattern has now become the most popular in RPL, factors can result in a trade-off in the network performance. As a matter
which is used to construct upward routing by broadcasting DIO mes­ of fact, OF may be a single metric such as ETX, energy, along with others,
sages across the DOADG. On the other hand, P2P and P2MP traffic or composite metrics (Darabkh et al., 2021b). To prolong the network
patterns transmit data along the pre-established DODAG during down­ life - time and achieve reliability in WSNs, a node’s residual energy and
ward routing by transmitting DAO messages on the reverse routes. As a LQ are integrated together to develop optimal metrics.
result, RPL is unable to take advantage of proper routing support in both The authors of (Kamgueu et al., 2015) presented a novel viewpoint
directions on the same network. Thus, difficulties with longer delays and on OF design. They suggested the implementation of the FIS for the first
lossy connectivity may arise specifically when the P2P traffic pattern is time in the RPL. It was mostly specified for ambiguous systems. The FIS
employed, which is a fundamental requirement for several LLNs appli­ is capable of merging, reasonably, several OFs into one. Moreover, the
cations. In non-storing mode, the root node must act as an intermediary. possibility of multi-parents, in highly dense networks, can be further
Consequently, the root node easily becomes a bottleneck when traffic is investigated. Once the traffic load in the multicast traffic is balanced, or
high. As stated earlier, downwards routes are utilized to support the the parent switching design in the traffic routing is proposed, the
P2MP traffic pattern in a DODAG. Nevertheless, they may not be the network load is lessened, and thus the network lifetime is prolonged.
finest quality option because of the asymmetric nature of wireless
channels in LLNs, which influences the performance of routing pro­ 6.4. QoS challenge
tocols. Protocols that do not take into account link asymmetry fail when
they encounter asymmetric links in real-world deployment (Misra et al., Reliable data transmission is essential for many LLN applications.
2011). A complementary IETF standard protocol (Goyal et al., 2013) has Implementing QoS features such as decreasing packet loss ratio, opti­
been suggested to solve the congestion and latency problems in the P2P mizing throughput, extending network lifetime, lowering energy

25
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

consumption, and avoiding excessive delays can help achieve this. consumption with low duty cycles. The dynamic duty cycling has been
Adopting high QoS necessitates improved routing decisions, enhanced shown to have a considerable impact on network layer protocols’ per­
transmission rates, and effective topological repair (Kharrufa et al., formance, causing long latency, low throughput, and low packet de­
2019). livery ratio (Sun et al., 2009). The features of other layer protocols
should be considered while designing routing protocols based on RPL for
6.5. Mobility challenge LLNs. As a result, cross-layer design is necessary to improve RPL per­
formance (Zhao et al., 2017) (Witwit and Idrees, 2018). Also, BLE is a
One obvious drawback to using the RPL protocol is that it lacks popular low-power and low-cost wireless protocol (Lee et al., 2016).
mobility support. Therefore, several researchers have paid attention to The cross-layer issue mainly concerns the discrepancy between the
finding solutions to accommodate mobile nodes. Mobile nodes play a payload at the MAC layer and the network layer. Since the MTU of an
crucial role in most IoT applications, particularly for urban and indus­ IEEE 802.15.4 MAC is 127 bytes and an IPv6 network MTU is 1280
trial applications (Perkins et al., 2011). Consequently, an efficient bytes, an adaptive layer (i.e., 6LoWPAN) is required for dealing with
routing protocol for mobile nodes is required to provide fast and reliable data packet fragmentation and reassembly as well as header compres­
communication between mobile and static nodes (Palattella et al., sion. It has been demonstrated in (Gardasevic et al., 2015) that as the
2013). Mobile nodes tend to cause dynamic changes in the network UDP payload increases, routing performance, such as delay and packet
topology and link failures. The emerging practical issue is how to deal delivery ratio, will deteriorate in both unicast and multicast situations.
with the sudden change of the preferred parent. This piece of informa­ Hence, increasing the packet payload will affect the routing perfor­
tion has a considerable influence on the network’s reliability and sta­ mance by consuming more energy, reducing packet delivery ratio, and
bility. Many efforts are being made to improve the enhanced versions of increasing end-to-end delay. The challenge of how to modify the routing
RPL associated with the mobility issue. strategy based on the packet payload needs to be further explored (Liu
et al., 2017). As a matter of fact, every layer in the IoT architecture has a
6.6. Workload balancing mechanism for RPL challenge set of responsibilities that directly affect the node’s energy consumption.
Moreover, each layer has a set of energy-efficient procedures to maxi­
Due to heterogeneous traffic patterns and random deployment of mize the network lifetime. However, the energy depletion of a node in
nodes, an unbalanced distribution of workloads occurs. In general terms, LLNs can cause serious problems for some applications, especially if a
in data-gathering applications, nodes adjacent to the root tend to node in the network becomes detached. Indeed, this necessitates
consume their energy faster than other nodes. As a result, the energy meeting long-term communication needs, determining and scheduling
hole problem will emerge, causing network performance to suffer (Liu medium access, and taking advantage of frequency diversity and chan­
et al., 2013a). The data gathering method is required to identify a so­ nel hopping. To use these strategies, we need precise synchronization
lution to the problem with the desired reliability (Jun et al., 2015). across all nodes in LLNs. Remarkably, effective cross-layer designs are
Workload sharing is also an effective approach for addressing energy required for low duty cycles and high throughput, especially at the lower
holes and maintaining network stability. layers.

6.7. Congestion detection and control challenge 6.9. Security challenge

The standard RPL did not define a well-mannered mechanism for The standard RPL does not identify the security mechanism. Thus,
dealing with congestion. Hence, it is essential to provide reliable RPL does not guarantee network security during different stages, i.e.,
congestion detection and control approaches to mitigate the congestion joining the DODAG or routing repair procedures (Verma and Ranga,
issue and enhance the RPL performance. Congestion is one of the sub­ 2020a). In particular, the data is prone to being lost, and the packets are
stantial issues in multi-hop routing in LLNs since the number of hops abandoned. As a result, attacking the RPL is simple (Avila et al., 2020).
raises the accumulated data, which causes congestion, especially at both This is considered critical for LLN applications whose main objectives
the node and link levels (Kharrufa et al., 2019). As mentioned earlier, are security and privacy, such as healthcare applications, which carry
nodes usually have a limited buffer size, triggering them to flood under highly individual and confidential data that cannot be shared with
high traffic, particularly with burst traffic. In other words, when mul­ others (Almusaylim et al., 2020). Thus, security is urgently needed to
tiple nodes transmit at high rates, the node buffer and wireless channel design novel security RPL-based protocols. This prompted us to explore
become congested (Al-Kashoash et al., 2017a). Furthermore, congestion and present a comprehensive study of RPL’s various security issues.
degrades the channel quality and increases the packet loss ratio, Before we get into the key RPL security challenges, it is worth noting
resulting in increased energy consumption, long delays, channel that protecting customer privacy necessitates encrypting data and
contention, packet drops, and lower reliability. Various approaches have transmitting it as cipher-text. Despite this, the encryption techniques
been suggested in the literature to alleviate the congestion problem, the will increase data traffic, control message overhead, energy consump­
most popular of which are resource control, traffic control, and hybrid tion, and end-to-end delay, making this process inappropriate for LLNs,
models. However, without taking into account the lossy nature of LLNs, which contain constrained devices. As a result, this survey provides
these approaches cannot provide an effectual routing (Zhao et al., 2017). readers with a comprehensive understanding of how to design and
develop new proposals that consider the aforementioned network per­
6.8. Cross-layer issue challenge formance metrics in order to make a trade-off between them. RPL se­
curity is threatened by the nature and constraints of LLNs (Adat and
The RPL protocol is compatible with a wide range of MAC and PHY Gupta, 2018). RPL is subject to a variety of routing attacks, which can be
protocols, particularly IEEE 802.15.4. Because MAC and PHY layer pa­ divided into two types: WSN-inherited attacks and RPL-specific attacks.
rameters have immediate effects on energy usage and link reliability, a Moreover, these attacks can be classified into outsider and insider at­
cross-layer approach to incorporating low-layer elements into metrics tacks. An outsider attack is carried out by a node that is not connected to
design could provide additional routing benefits (Liu et al., 2017). In the established network, while an insider attack is performed by a node
fact, IEEE 802.15.4 is designed to prevent interference and multipath that is connected to the established network (Homoliak et al., 2019)
fading and operates at extremely low power. It is the ideal candidate for (Abomhara and Køien, 2015).
the MAC layer protocols (Cuomo et al., 2013). Other notable features of Not to this extent, but rather, the authors in (Mayzaud et al., 2016)
IEEE 802.15.4 include the use of the CSMA-collision avoidance scheme, provided a classification of routing attacks against the RPL protocol.
the allocation of guaranteed time slots, and exceptionally low power This one considers the attack’s targets, as well as which parts of the RPL

26
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 7
Various RPL security attacks and their effects on the network.
Attack Type of attack The reason of incidence Negative repercussions for network Ref.
performance

Increased Rank Insider RPL-based Resources DIO − An attacker can modify a node’s rank − The network performance is Shafique
attack or change the parent selection degraded in terms of end-to-end et al. (2018)
Decreased Rank Insider RPL-based Traffic DIO algorithm to undertake the worst delay, packet delivery ratio, energy
attack parent selection attack. consumption, congestion, and
Worst parent attack Insider RPL-based Topology DIO − The rank field or rules can be abused. control overheard.
− Generating loops and inefficient
routes.
Flooding attack Insider/ Inherited- Resources DIO/ − An attacker can modify a node’s rank − The control overhead, congestion, Verma and
Outsider WSNs DIS or change the parent selection packet loss ratio, energy Ranga
algorithm to undertake the worst consumption, end-to-end delay, (2020b)
parent selection attack. and node saturation are increased.
Routing table Insider RPL-based Resources DAO − An attacker transmits many fake − The victim node is unable to Kamble et al.
flooding attack routes using DAO messages to flood construct proper routes. (2017)
the victim’s routing table. As a result, − The energy consumption and
the vulnerable node ignores receiving memory usage are increased.
legitimate DAO messages.
Selective Insider Inherited- Topology DIO − An attacker prevented or discard − Increase end-to-end delay and Airehrour
forwarding (grey WSNs some packets to be transmitted by a packet loss ratio. et al. (2017)
hole) attack malicious node. − Decrease packet delivery ratio.
− Interrupt the routing path and
isolate a part of the network.
Blackhole attack Insider Inherited- Topology DIO − A malicious node discards all packets − Interrupt the routing path and Airehrour
WSNs as they pass through it. isolate a part of the network. et al. (2016)
− Increase packet loss ratio and end-
to-end delay.
Wormhole attack Insider Inherited- Topology DIO − Two or more attackers in the network − Establish inefficient routes. Khan et al.
WSNs establish a high-bandwidth tunnel (2013)
between them, advertise these tun­
nels as high-quality routes.
Sinkhole attack Insider Inherited- Topology DIO − An attacker node seeks to attract the − Establish inefficient and non- Parkavi and
WSNs most traffic by decreasing its rank to optimized routes. Arockiam
become the preferred parent, − Degrade the network performance. (2021)
allowing it to control the majority of − Increase end-to-end delay.
the data traveling through the
network.
DODAG Version Insider RPL-based Resources DIO − An attacker can modify the Version − Shorten the network’s lifespan Aris et al.
Number attack Number field in a DIO message and since the nodes waste energy (2016)
broadcast it to the neighbor nodes. reconstructing the network
This results in a global repair, which topology unnecessarily as a result
leads to the needless rebuilding of the of the attack. Also, decrease the
whole DODAG. packet delivery ratio.
− Raise the end-to-end delay and
control overhead.
− Generate routing loops and rank
variations.
Neighbor attack Insider RPL-based Topology DIO − An attacker transmits an unmodified − Increase end-to-end delay. Le et al.
DIO message to its surrounding (2013)
nodes.
DODAG Insider/ RPL-based Resources DIO − An attacker misuses R and O flags of − Increase control overhead, energy Sehgal et al.
inconsistency Outsider IPv6 header options (these flags are consumption, packet loss ratio, and (2014)
attack related to detecting inconsistency traffic congestion due to
(loops) and starting the repair unnecessary repairs.
process) to perform various attacks.
The R flag is set to 1, so the trickle
timer is reset.
Routing choice Insider RPL-based Resources DIO − An attacker learns the current routing − Establish inefficient routes. Zhang et al.
intrusion attack settings that nodes use to choose the − Increase control overheads, loops, (2015)
best routes. The DIO messages are energy consumption, and end-to-
then captured, and the faked DIO end delay.
messages are later broadcast using the
legitimate identity. Also, this can
modify the routes.
DIO suppression Insider/ RPL-based Topology DIO − An attacker suppresses the victim − Unoptimized routes are created, Perazzo et al.
attack Outsider node to send DIO messages required and the network is portioned. (2017)
to construct the network or remove
stale paths.
DIS attack Insider/ RPL-based Resources DIS − An attacker sends periodic DIS − Increase control, overhead, energy Le et al.
Outsider messages to the victim nodes, so they consumption, and end-to-end (2016)
are imposed to reset their trickle delay
timers, respond with new DIO − Creating inefficient routing paths.
messages, and they are flooded.
ETX manipulation Insider RPL-based Resources DIO − An attacker tries to control the ETX − Creating inefficient routing paths. Shreenivas
values of nodes to obtain lower values et al. (2017)
(continued on next page)

27
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 7 (continued )
Attack Type of attack The reason of incidence Negative repercussions for network Ref.
performance

to direct the network traffic through it

and attain a better location in the
network.
Sybil or Identity Insider Inherited- Traffic DIS − An attacker node establishes several − Increase energy consumption and Zhang et al.
attack WSNs identities and tricks the network control overhead. (2014)
nodes. Thus, the attacker node can − Lessen packet delivery ratio.
manage many regions of the network − Sensitive data is subject to thief.
without employing any physical
nodes.
Replay attack Insider RPL-based Topology DIS/ − An attacker keeps track of valid − Nodes update their routing tables Le et al.
DIO/ control messages from other nodes with old data as a result of replay (2013)
DAO/ and forwards them to other nodes in attacks, leading to a fake topology.
DAO- the network later.
ACK
Falsification of Insider RPL-based Topology DAO − An attacker changes DAO message − Establish wrong and fake routes Mayzaud
routing table fields to construct fake routes by − Increase end-to-end delay packet et al. (2016)
attack filling the routing table of the victim loss ratio and traffic congestion.
nodes with false routing information,
particularly in the storing mode.
Local repair attack Insider RPL-based Resources DAO − An attacker forces the victim nodes to − Increase end-to-end delay, energy Pongle and
start needless local repairs by consumption, and control Chavan
changing the rank to infinity or overhead. (2015)
modifying the DODAGID, although
the nodes are connected to their
parent.
DAO inconsistency Insider RPL-based Topology DAO − An attacker transmits a fake DAO − Increase end-to-end delay. Pu (2018)
attack message, which contains fake routing − Formation of unoptimized network
information, Thus, a node’s child topology by generating long routes
node is unable to send data to a or isolation of nodes.
specified destination owing to the
absence of a route learned from a fake
DAO message during topology
formation. By changing the “F" flag to
1 in the packets and sending them
back to their parent, the attacker
exploits this process to carry out an
attack. As a result, the parent node is
forced to remove valid downward
routes.
Fake Insider RPL-based Topology DAO- − An attacker can send a fake − Increase end-to-end delay and Bang et al.
acknowledgment ACK acknowledgment in response to the control overhead. (2022)
attack packet received, but the attacker can
discard it. As a result, this attack may
increase end-to-end delay and
network overhead.

network, are affected. Because of that, they divided security attacks into 6.10. Authors’ insights
three groups. The first group includes attacks focused on reducing
network resources (memory and energy). Such attacks are especially RPL is widely known as the standard routing protocol for the IoT. It is
devastating to these constrained networks since they drastically reduce stated by the IETF to meet the requirements of LLNs. After investigating
the node lifetime and, hence, the RPL network’s lifetime. Attacks against RPL from all aspects, we have observed that it has many drawbacks and
the RPL network topology fall under the second group. They impair the limitations that impact its function. Some of these challenges stem from
network’s basic operation: the topology may be sub-optimized relative its functionalism, such as obtaining efficient routing, which is associated
to normal network convergence, or a group of RPL nodes may be with the inefficacious downward mechanism, trickle algorithm, and
disconnected from the network. Finally, attacks on network traffic, such multicast strategies. Some limitations come into view with the ubiqui­
as eavesdropping and misappropriation attacks, are the third category. tous IoT-LLNs applications that have various requirements. These limi­
Additionally, the authors in (Perkins et al., 2011) presented an tations are mainly energy usage and diversification of OFs, which may
up-to-date survey concerning security. Specifically, they presented a be a single metric or multiple metrics. Also, the emergence of applica­
novel classification of RPL security attacks based on the type of control tions that require the moving of sensors, such as health applications,
messages (i.e., DIS, DIO, DAO, and DAO-ACK). The literature identifies necessitates a well-mannered mobility algorithm. Along with that, QoS
various types of attacks that can emerge while utilizing the RPL routing considers the main concerns in such applications, including reliability,
protocol. The main difference between us and them is that our survey end-to-end delay, control overhead, and memory usage. Additionally,
attempts to assign a different type for each attack. Thus, it will be load balancing, as well as congestion detection and control, must be
concise and easy to understand in order to appeal to the widest possible addressed, especially as the network grows larger.
audience. Table 7 summarizes the major types of attacks in the first Not to this extent, but rather, applying a new cross-layer design by
column. The second, third, fourth, and fifth columns classify the types of changing the IETF IPv6 stack is a hot research area and encourages the
attacks based on the other surveys’ categorization. The reason for the designers to put more effort into implementing a new stack with
attack is shown in the seventh column, and the eighth column indicates different MAC and PHY layers rather than the IEEE 802.15.4. Further­
the negative impact of the attack on the network performance metrics. more, security is a very critical factor in mankind’s lives, and everyone
searches for the privacy of their lives, thus any attack may endanger

28
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Fig. 18. Taxonomy of enhanced routing solutions for RPL protocol.

their lives or businesses. However, the standard RPL has not defined the P2P-RPL is that it overflows the network with control messages,
security mechanism. In the end, the classification of the above- increasing the overhead and energy consumption. Many researchers
mentioned challenges that we come up with is the result of our exten­ have studied multicast routing approaches to enhance the reliability of
sive research in greater detail. These findings will make the research the network. These researches suggested SMRF (Oikonomou and Phil­
process easier for readers, particularly those who are novices to RPL, to lips, 2012) (Oikonomou et al., 2013), ESMRF (Abdel Fadeel and El
save time and effort. Sayed, 2015), and BMRF (Gastón Lorente et al., 2017) to manage RPL’s
multicast transmission messages. The experimental outcomes demon­
7. Enhanced routing solutions for RPL protocol strate that these protocols can upgrade the trickle algorithm. Besides,
they ensure high reliability by using link-layer broadcast and link-layer
Many researchers have focused on RPL, the most ubiquitous LLN unicast. However, the results show up the degradation of network per­
routing protocol, aiming to struggle with routing challenges, including formance in terms of energy consumption and end-to-end delay. The
efficient routing support, energy consumption, diversification of OFs, authors in (Oikonomou and Phillips, 2012) and (Oikonomou et al.,
mobility, load balancing, congestion detection and control, cross-layer 2013) have proposed SMRF. It used the “storing mode of operation with
issues, and QoS. Accordingly, various enhancement solutions have multicast support” of the RPL standard. SMRF proposed a cross-layer
been developed for RPL, and they are classified based on their proposals approach that enhances the operation of RDC protocols for multicast
and challenges, as displayed in Fig. 18. The subsections that follow show transmitting processes. Multicast, in RPL, does not define any control
all of the solutions that were presented. The aims, techniques, strengths, procedure to sidestep the receiving and forwarding of duplicated mul­
and limitations of each strategy are listed in a summary table following ticast packets. By this means, the node in SMRF is obliged only to handle
each subsection. The aims of the upgraded routing solution are specified multicast packets received from its preferred parents. Nevertheless,
in the “Aim” column. The “Strategies” column illustrates the proposal’s SMRF has limitations when compared to native RPL multicast. The
approach to achieving its goals. Ultimately, the strengths and limitations weakness of SMRF is that it does not support the upward with multicast
columns highlighted the significant contributions and restrictions of the technique efficiently.
suggested routing approach, respectively. To overcome the SMRF drawback, the ESMRF was proposed in
(Abdel Fadeel and El Sayed, 2015). In this approach, the multicast
messages can be transmitted in both directions: upward and downward.
7.1. Aefficient routing support enhancements ESMRF suggested that each multicast message be sent to the root node
first. Then, the root should forward the multicast message to its desti­
In (Baccelli et al., 2011), the authors proposed an expanded version nation. Accordingly, when a node transmits a multicast message, it has
of RPL, namely P2P-RPL. It is a reactive route discovery mechanism to compress its content and the destinations in the unicast ICMPv6
intended to be the vital communication model for the traffic patterns in delegation packet. This message is communicated with all other nodes.
LLNs applications. P2P communication has drawn more attention as a The body of the multicast message is extracted by the root node once the
result of its widespread and crucial usage in real-life. Several IoT ap­ packet is received, and the message is then directed to the destination.
plications, including building automation applications and However, since the root node has the role of transmitting overall mul­
remote-control applications, need a provision of a well-organized P2P ticast messages, this approach is too costly in large networks, as it in­
routing protocol. Unlike RPL, P2P-RPL finds the best-quality path for creases the latency and the communication overhead. The authors, in
any source-destination pair (Zhao et al., 2017). The main drawback of

29
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

(Ko et al., 2015), proposed the DualMOP-RPL. It concentrates on the standard as it has a lower end-to-end delay. Accordingly, the packet
interoperability issue among the downward routing modes. Also, it delivery ratio has improved in both modes. The shortcoming of the
tackles the connectivity issue in a single DODAG, which holds nodes hybrid mode RPL is that it does not support mobility.
operating in either storing or non-storing mode. DualMOP-RPL attempts Trickle algorithm optimization is another way to maintain an effi­
to solve this issue by configuring the router node to work in both modes cient route in RPL. Actually, the trickle algorithm is identified to regu­
simultaneously. Therefore, it stores routing information in storing mode late the time of dissemination of the DIO messages to keep the network
while it manages the source-routing if the non-storing mode is applied. consistent. Nevertheless, it has some limitations that impact the network
As a result, this enhancement does not influence the network perfor­ performance in terms of convergence time, energy consumption, and
mance while the usage of computational resource devices is optimized. control overhead. These limitations motivated the authors in (Lamaazi
Furthermore, in (Zhao et al., 2016), ER-RPL was introduced. This and Benamar, 2019) to develop a novel RPL enhancement based on a
solution aims to provide P2P communication with efficient energy flexible trickle algorithm, namely RPL-FL. It determines three issues
without affecting the reliability of the network. This approach is based concerning the trickle algorithm. The most prominent one is the short
on partitioning the concerned network into diverse zones and attaining listen-only period, which has a low value of the Imin parameter. This
P2P route discovery, depending on every node’s location. P2P commu­ condition permits the nodes to transmit more control messages to make
nication is accomplished by transmitting a P2P route request message the convergence time of the network minimal. In view of that, increasing
from the source node towards the destination node, passing through the the number of control messages leads the nodes to consume more en­
root node. Afterwards, the destination checks whether the route is ergy. RPL-FL had addressed these problems and resolved them. First, it
adequate in terms of cost. When the answer is yes, the destination in­ reduces the time of sending the control message by fixing the trans­
structs, via the reverse route, the source to begin transmitting data. mission time T at I/2 rather than taking it randomly in [I, I/2]. Second, it
Otherwise, the destination starts the route discovery mechanism based specifies a high value of Imin to minimize the control overhead. Finally,
on region. The ER-RPL operation is complicated compared with the it implements three intervals to downgrade the energy consumption,
standard RPL. However, it reduces energy consumption by evading the which are the start of the process interval, the skipped interval, and the
network overflow with P2P route discovery messages, which are sent to end of the skipped process interval. The results proved that RPL-FL
other nodes without referring to the root node. achieves its target regarding energy consumption, packet delivery
Moreover, the authors in (Gastón Lorente et al., 2017) proposed a ratio, and control overheard. On the other hand, it does not consider
new multicast approach called BMRF. It combines the best characteris­ mobility.
tics of both RPL and SMRF. Consequently, three packet-transmitting RPL provides three traffic patterns (i.e., P2P, MP2P, and P2MP).
modes are used, including (i) Unicast mode: the sender node checks P2MP is multicast from the root to the nodes. However, the multicast
the nodes intended for the multicast message, then the message is sent mechanism in RPL is still an unresolved issue. This prompted the authors
individually in unicast style to the preferred parent. (ii) Broadcast mode: of (Hwang et al., 2020) to propose proactive multicast forwarding with
the packet sender checks that the intended nodes receive the multicast RPL (PMFR), which is based on the W-SPH) algorithm and is used to
message; thus, the communication among nodes is operated as a construct multicast topologies based on RPL message information while
broadcast. (iii) Mixed-T-mode, in which the sender sends packets in keeping the transmission nature of WSNs in mind. PMFR utilizes a
response to a threshold (T) that switches between multiple unicasts and multicast mechanism with proactive retransmission to enhance the
broadcasts. Hence, BMRF permits nodes to transmit and receive multi­ reliability of multicast packet transmissions. They used EMT as a routing
cast messages apart from their location in the network. Moreover, BMRF metric. This metric calculates the expected number of multicast
avoids disordered delivery. The authors in (Kim et al., 2017b) suggested retransmissions. Therefore, PMFR lets the packets be retrieved via
a way to enhance traffic patterns supported by RPL, named DT-RPL. This retransmission and boosts the opportunity of the nodes to successfully
suggestion aims to stabilize bidirectional communication throughout obtain those packets. The simulation findings confirmed that the PMFR
the network nodes. During packets’ transmission, in downward routes, outperforms multicast protocol for LLNs (Hui and Kelsey, 2016) and
the nodes examine the received messages to deliver LQ information. ESMRF in terms of energy consumption, packet delivery ratio, and
Then, the recipient node computes and updates the LQ information. On end-to-end delay. Table 8 summarizes RPL enhancements for the effi­
the other hand, RPL refreshes LQ information only during the upward cient routing support challenge.
route. The approach improves the process of building the routing table,
eschews the frequent triggering of global repair across the forwarding of 7.2. Energy consumption enhancements
downward messages, and enriches the packet delivery ratio. However,
DT-RPL increases communication overhead. The authors, in (Abreu et al., 2014), introduced the EAOF. It pro­
To suppress the dynamic-load issue in LLNs, the authors in (Taghi­ poses a solution to improve the network’s reliability and the lifetime of
zadeh et al., 2018) introduced a novel routing protocol called CL-RPL. biomedical WSNs, in particular for health applications. Wherein, smaller
The mechanism of CL-RPL is based on the LQ and energy information, nodes are required to work for long time periods without human inter­
as well as the workload of the routes. It aims to alleviate the packet loss ference and provide, at the same time, convenient levels for both reli­
ratio, extend the network lifetime in LLNs, intensify productivity, and ability and QoS. The proposed solution uses ETX and residual energy as a
highly change traffic. Consequently, the proposal reduces energy con­ routing metric on each sensor node to calculate the optimal routes to
sumption and the rate of packet loss. In (Oh et al., 2018), the authors transmit data packets across the network. Experiment results prove that
proposed a hybrid mode of RPL. It merges the features of both existing the EAOF prolongs the network lifetime and reduces energy consump­
modes in RPL to enhance the performance of downward-message for­ tion. In (Capone et al., 2014b), the authors proposed the L2AM. It is
warding. Besides, it provides a solution to the scalability issue, which is defined for two purposes: First, it balances energy consumption.
caused by the resource limitation of the node if the storing mode is Consequently, each node tends to deplete the same amount of energy,
utilized. On the other hand, in non-storing mode, the new proposal prolonging the network lifetime. Second, to ensure reliable data among
overcomes the growth in fragmentation stemming from routing over­ the routes, it utilizes ETX as a routing metric. Hence, the proposed
head. This approach introduces a novel routing header format. The metric permits the node to select routes that are reliable and
mechanism is as follows: the routing information for packet delivery is energy-efficient. Simulation results clearly show that L2AM prolongs the
disseminated with the extended routing header to a DODAG root, in network lifetime, given the same overall degree of network reliability.
addition to other nodes. The experiment’s outcomes prove that this The authors in (Belghachi and Feham, 2015) introduced a new
mechanism outperforms the storing mode in the default RPL since it has approach named: QoS_RPL. The authors suggest using new routing
fewer routing entries, and it surpasses the non-storing mode in the RPL metrics, including the transmission delay and the residual energy, in the

30
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 8
Summary table for RPL enhancements for efficient routing support challenge.
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

P2P-RPL 2011 - Improve P2P in RPL. - Build new reactive P2P - Enhance the - Increase energy consumption. Baccelli et al.
paths. performance in home and - Flooding the network with control (2011)
building automation messages.
scenarios. - Raise the communication overhead.
- Increase the packet
delivery ratio and lessen
the delay.
- Provide shorter paths by
avoiding going through
the root node.
- Reduce the bottleneck
at the DODAG root.
SMRF 2013 - Improve multicast data - Design a cross-layer - Achieve significant - Support only downwards multicast Oikonomou
transmission. approach that enhances RPL improvement in terms of traffic within a DODAG (the absence et al. (2013)
- Adopt a cross-layer approach. for multicast transmission. end-to-end delay and of upward multicast support).
- Design a trickle multicast energy efficiency. - Only messages sent from the node’s
algorithm. - Reduce packet delivery preferred parent are handled while
ratio. messages received from child nodes
- It does not require geo- are not handled or retransmitted as a
location information. result.
- Low memory
requirements and code
size.
- Evade duplicated
multicast packets
process.
ESMRF 2015 - Allow nodes to be able to - Multicast traffic sources - Support both upward - Emerge a bottleneck at the root Abdel Fadeel
send multicast traffic upward encapsulate their multicast and downward traffic for node. and El Sayed
and downward. packet and send it to the root multicast data. - High end-to-end latency. (2015)
that forwards a multicast - Improve the reliability - Increase in energy consumption.
packet. of the network.
- Improve the packet
delivery ratio.
DualMOP- 2015 - Support nodes with different -Modify RPL’s control packets. - Upgrade the link - Complicate the process of control Ko et al. (2015)
RPL MOPs for downward routing to - Provide different MOPs at the qualities. messages
communicate gracefully in a DODOG tree. - Enhance the network’ - Not compatible with native RPL.
single RPL network. routing performance.
- Permit for network
migrations and an
alternate routing in the
case of root failures.
ER-RPL 2016 - Support generic traffic - A hybrid of a proactive and - Achieve efficient energy - Nodes need to be aware of other Zhao et al.
patterns. reactive routing protocol. of data delivery without nodes’ locations. (2016)
- Achieve reliability and - Use of the region’s influencing the - Complex approach compared to the
efficient energy information of networks. reliability. default RPL.
simultaneously. - Reduce routing - Produce a new control message
overhead. besides the native RPL control
- Improve the packet messages.
delivery ratio.
BMRF 2017 - Enhance SMRF’s multicast - Combine the best features of - Lessen radio - Increase memory usage. Gastón Lorente
data forwarding. RPL multicast and SMRF. transmissions and energy - Increase end-to-end delay. et al. (2017)
- Offer a choice between Link consumption.
Layer broadcast and Link - Higher packet delivery
Layer unicast depending on a ratio.
threshold value. - Improve reliability.
DT-RPL 2017 - Refresh routes in dynamic - Use both upward and - Improve the packet - Does not support large scale Kim et al.
link environments regardless downward. packets for LQ delivery ratio. networks. (2017b)
of traffic patterns. updates. - Mitigate control
overhead and energy
consumption.
CL-RPL 2018 - Overcome the problems of - Consider the status of a - Improve network - Slightly higher memory usage. Taghizadeh
RPL under heavy and dynamic parent chain. lifetime. - Raise the end-to-end delay. et al. (2018)
load networks. - Consider the remaining - Decrease packet loss
queue and the energy level of ratio.
candidate parents. - Shorten energy
- Use ETX as a routing metric. consumption.
Hybrid 2018 - Improve the performance of - Store routing information by - Increase the packet - No mobility support. Oh et al. (2018)
mode- storing and non-storing modes disseminating this information delivery ratio.
RPL in downward routing. to the DODAG root and - In the storing mode, it
routers. has fewer routing entries.
- In the non-storing
mode, it has a lower end-
to-end delay.
(continued on next page)

31
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 8 (continued )
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

RPL-FL 2019 - Develop a flexible trickle - Fix the transmission time T at - Improve the packet - No mobility support. Lamaazi and
algorithm. I/2. delivery ratio, energy - Does not work with hybrid RPL Benamar
- Specify a high value of Imin. consumption, and control improvements. (2019)
- Implement three intervals to overhead. Lamaazi and
downgrade the energy Benamar
consumption. (2020b)
PMFR 2020 - Provide a reliable multicast - Utilize EMT as a routing - Increase energy - There is no mobility support. Hwang et al.
mechanism metric. efficiency and packet (2020)
- Employ W-SPH algorithm delivery ratio.
that is used to construct - Lessen the end-to-end
multicast topologies. delay.

next-hop selection mechanism for the RPL protocol. An OF is designed, Subsequently, the network lifetime is extended. The main downside of
for this metric, based on ACO, which seeks to better achieve the re­ this approach is that each application requires a specific OF, and the
quirements of efficient energy and QoS in LLNs. The obtained outcomes suggested proposal does not support mobility.
have revealed that the proposed solution outperforms the performance The authors in (Lalani et al., 2020) studied the effect of mobile nodes
of the standard RPL in terms of energy depletion and network lifetime. on the network performance since the standard RPL does not support
Additionally, QoS_RPL distributes the energy uniformly between nodes. mobility. Hence, it does not provide up-to-date information about the
The authors (Iova et al., 2015) suggested an energy-balancing routing new parents. Furthermore, in a mobility scenario, the neighbor table still
protocol, which prolongs the nodes’ lifetimes. This approach is called comprises parents who are out of date. They introduced REFER, which is
multiELT-RPL. They proposed the ELT metric, representing the a reliable and energy-efficient RPL-based protocol for mobile applica­
remaining time of a node (the time at which the node runs out of en­ tions. It uses a new neighbor placement mechanism based on the leasing
ergy). Also, they proposed a multipath approach to evade network time of the parents and takes into account several node/link metrics to
instability and convergence issues. Furthermore, a mechanism has been choose the preferred parent to acquire the optimal path to the destina­
introduced to limit the switching of parents. The simulation findings tion. PEFER uses a combination of metrics, for instance, ETX, RSSI, the
show that both routing reliability and network lifetime are improved distance between two nodes, rank, movement status, and ELT. PEFER
while the number of DODAG reconfigurations is minimized. As well, the chooses the preferred parent based on the priority of these metrics.
overhead of control messages and energy consumption are reduced. According to the experimental results, PEFER minimizes energy con­
On the other hand, the authors in (Khelifi et al., 2015) presented the sumption while also improving network stability.
Pro-RPL. It is an extended version of RPL. It utilizes a proactive approach Finally, the authors in (Pushpalatha et al., 2021) examined the
to determine failures in the networks. This scheme exploits a “suffering behavior of the ETX metric for lengthy routes and the effect of the node
index” that reflects the cost of network failures. This protocol aims to or link failures on those routes. Their study realized that long routes fail
predict the occurrence of a problem before it actually occurs and triggers due to the scarcity of information from neighboring links. As a result,
the local repair mechanism as designed in the standard of RPL. Thus, it they made great efforts to establish stable routes by suggesting a new
minimizes failures before they influence the network connectivity. This protocol named L-RPL. It is a Laplacian energy-based path selection
approach is not complex and does not consume much memory to pro­ protocol that advances RPL by including additional information about
cess. The experimental outcomes indicate that this proposal reduces the neighbor connections in the path selection process. Along with ETX,
packet loss ratio. Besides, it prolongs the network lifetime by minifying they discovered a new metric for link estimation named Laplacian en­
energy consumption. In (Zhao et al., 2016), the concept of ER-RPL was ergy drop, which is a two-walk neighborhood connectivity measure
proposed. All information about this protocol can be found in the effi­ coupled with L-RPL to enable the node to choose the most robust path. In
cient routing support enhancements section. The objectives of SSCF, light of that, the links in the upper 2-walk paths are evaluated by
which was developed in (Khan et al., 2016), are to extend the network’s computing the Laplacian energy drop of the parents. Accordingly, the
lifetime and increase throughput. Furthermore, in SSCF, the network paths that have resistant connectivity are selected. The L-RPL calculates
status seen at a sink is exchanged with its neighboring sinks via periodic the Laplacian energy drop using a procedure where each node computes
route maintenance messages issued by RPL. SSCF processes specify the its Laplacian energy drop and broadcasts it along the same lines as the
syntax and semantics for exchanging network optimization parameters rank metric. The child nodes broadcast the summation of the link
between sink nodes. To fulfill the throughput optimization, the sink weights of their parents. The simulation results found that L-RPL en­
nodes, in SSCF, set their network size for better load balancing. The hances the network performance in terms of end-to-end delay, control
simulation findings prove that the new approach improves throughput overhead, efficient energy use, and reliability. Crucially, the authors of
as well as reduces energy consumption. Therefore, SSCF prolongs the (Sennan et al., 2021) concentrated on introducing a new RPL-based
network lifetime. Besides, the packet delivery ratio has increased. protocol that reduces network energy consumption. To achieve this
The authors in (Araújo et al., 2018) proposed a new design to select goal, they utilized two techniques, which include clustering and a new
routes using the fuzzy logic system named: DQCA-OF. Four OFs are optimization method. Basically, LOA-RPL is made up of three steps, that
proposed and selected dynamically based on context information. OFs are, cluster formation, cluster head selection, and route establishment.
are generated from the amalgamation of the following routing metrics: The Euclidean distance is used to form a cluster, and the LOA is used
ETX, HC, and consumed energy. Moreover, DQCA-OF also considers a further to select the cluster head. Finally, the route is established using
route classifier based on a fuzzy logic system by approximating the de­ the remaining energy information. Regarding the network performance
gree of quality of the routes. The route classifier operates as an input to metrics, which include packet delivery ratio, throughput, energy con­
the values designated for the routing metrics, guaranteeing consistency sumption, and network lifetime, LOA-RPL outperforms its counterparts.
in decision-making within the routing process. Simulation results have Actually, this protocol does not support mobility and should primarily
shown that DQCA-OF outperforms other protocols since it provides high consider security aspects because there is a colossal amount of data
reliability, reduces the end-to-end delay for data delivery, and improves generated from clustering. The studies, discussed in this subsection, are
the QoS of the network. Besides, it lowers energy consumption. summarized in Table 9.

32
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 9
Summary table for RPL enhancements for energy consumption challenge.
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

EAOF 2014 - Offer the appropriate levels - Use the ETX metric and the - Enlarge the network lifetime. - Minify packet delivery Abreu et al.
of reliability and QoS for residual energy to calculate the - Lessen the energy ratio. (2014)
efficient energy. optimal paths to route data packets consumption.
across the network.
L2AM 2014 - Aim to balance energy - Consider data reliability along the - Prolong the network lifetime. - No mobility support. Capone et al.
consumption. routes, as defined by the ETX - Provide reliability. (2014b)
routing metric. - Balance energy consumption.
- Nodes use routes that are reliable
and energy efficient.
QoS_RPL 2015 - Satisfy the requirements of - Utilize the remaining energy and - Lower the energy depletion. - Packet delivery ratio is Belghachi and
efficient energy and QoS in the transmission delay as routing - Increase the network lifetime. slightly decreased. Feham (2015)
LLNs. metrics for the next-hop selection. - Boost communication
- The OF is built using the (ACO) overhead messages.
method.
multiELT- 2015 - Reduce the energy - Use ELT as a routing metric. - Control overhead messages - Change control message Iova et al.
RPL consumption. - Provide a multipath approach. and energy consumption are structure. (2015)
- Maximize the lifetime of the - Use of a procedure to control the reduced. - No mobility support.
nodes and distribute traffic parental exchanges. - Minimize the number of - Raise the usage of
load. transmissions. memory slightly.
- Reduce the number of DODAG
reconfigurations.
Pro-RPL 2015 - Exploit a proactive approach - Use a suffering index that reflects - Pro-RPL is not a complex - No mobility support. Khelifi et al.
to establish failures and detect the cost of network failures. approach. - No practical (2015)
them. - Predict the problem before it - Does not consume much experiments.
occurs. memory.
- Trigger the local repair - Reduce the packet loss ratio
mechanism. and energy consumption.
- Prolong the network lifetime.
ER-RPL 2016 - Support generic traffic - A hybrid of a proactive and - Achieve efficient energy of - Nodes need to be aware Zhao et al.
patterns. reactive routing protocol. data delivery without of other nodes’ locations. (2016)
-Achieve reliability and - Use of the region’s information of influencing the reliability. - Complex approach
efficient energy. networks. - Reduce routing overhead. compared to the default
- Improve the packet delivery RPL.
ratio. - Generate a new control
message besides the
native RPL control
messages.
SSCF 2016 - Aim to increase the network’s - Sink-to-sink coordination - Higher throughput. - No practical Khan et al.
throughput and lifetime. framework. - Low energy consumption. experiments. (2016)
- Use multiple sinks. -Increase the network lifetime. - No mobility support.
- Improves the packet delivery - Incompatible with the
ratio. native RPL.
DQCA-OF 2018 - Fulfill the requirements of - Use the fuzzy logic system based on - High reliability and low delay. - Increase memory Araújo et al.
specific applications. ETX, HC, and the consumed energy - Increase the QoS and network consumption due to the (2018)
routing metrics to select the routes. lifetime. implantation of the fuzzy
- Reduce the use of energy. system.
- Each application
requires a specific OF. -
No mobility support.
- The change in the
number of sensor nodes
and RFID tags is not
considered.
PEFER 2020 - Support mobility. - Use a new neighbor placement - Upgrade the reliability of the - High memory usage. Lalani et al.
- Achieve reliability in mobile mechanism depends on the leasing network. (2020)
IoT applications. time of the parents. - Reduce energy consumption.
- Uses a combination of metrics, for
instance, ETX, RSSI, the distance
between two nodes, rank,
movement status, and ELT to select
the preferred parent.
L-RPL 2021 - Maintain global neighbor - Identify the Laplacian energy drop - Augment the network - Does not support Pushpalatha
connectivity information into metric. performance in terms of delay, mobility. et al. (2021)
path selection to deliver - Evaluate the links in the upper 2- control overhead, efficient - Heavy calculation
steady routes for lengthier walk paths by computing the energy, and reliability. reduces memory usage.
routes. Laplacian energy drop of the
parents.
- Select the paths that have resistant
connectivity.
LOA-RPL 2021 -Minimize duplicate data. - Exploit two techniques: clustering - Extend network lifetime. - Does not support Sennan et al.
- Save energy and prolong and the LOA optimization method. - Increase packet delivery mobility. (2021)
network lifetime. - Utilize Euclidean distance to form ration and throughput. - Needs to investigate
the cluster, LOA to select the cluster - Reduce the memory usage. security aspects.
head, and remaining energy to - Reduce energy consumption.
establish the route.

33
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

7.3. Diversification of OFs enhancements quality. The proposed scheme has been proven, by the simulations, to
have a significant improvement in the performance in terms of network
EAOF (Abreu et al., 2014) and L2AM (Capone et al., 2014b) are lifetime, end-to-end delay, and packet loss ratio. On top of that, OF-FL
mentioned in the previous subsection (energy consumption enhance­ has been designed with only minimal add-ons to make it compatible
ments) with more specific details. OF-FL was designed by the authors of with the native RPL. Another point of strength in this approach is that it
(Gaddour et al., 2014). The parameters of the fuzzy system are being supports several simultaneous applications with antagonistic re­
relied on to make configurable routing decisions by this novel OF using quirements. On the other hand, the weakness of this approach lies in the
specific routing metrics. In OF-FL, choosing the best paths to the desti­ fact that OF-FL has high memory usage when it comes to using fuzzy
nation depends on the application requirements. OF-FL handles the logic. Another design for using the fuzzy logic-based OF was presented
fuzzy logic system to utilize the following link and node routing metrics: in (Kamgueu et al., 2015). The proposed solution utilizes the FIS to
end-to-end delay, HC, ETX, and the level of battery energy. A set of combine the following routing metrics: ETX, delay, and the residual
IF-THEN rules are used to get the preferred parent, as this combines the energy of the node into one single value (quality), aiming to improve
four linguistic variables. The fuzzy rules are estimated such that the QoS and energy consumption. A real sensor network deployment is ar­
affiliation of the neighbor in the fuzzy subsets of neighbors is reverted to ranged in an indoor environment to evaluate the metric. In the proposed
the best qualities. The preferred parent is the neighbor with the best approach, nodes’ rank is obtained by attaining the sum of their preferred

Table 10
Summary table for RPL enhancements for the diversification of OFs challenge.
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

EAOF 2014 - Offer the appropriate levels - Use the ETX metric and the residual - Enlarge the network lifetime. - Minify packet delivery ratio. Abreu et al.
of reliability and QoS for energy to calculate the optimal paths - Lessen the energy (2014)
efficient energy. to route data packets across the consumption.
network.
L2AM 2014 - Aim to balance energy - Consider data reliability along the - Prolong the network - No mobility support. Capone et al.
consumption. routes, as defined by the ETX routing lifetime. (2014b)
metric. - Provide reliability.
- Nodes use routes that are reliable - Balance energy
and energy efficient. consumption.
OF-FL 2014 - Offer a configurable routing - Combine the following metrics: HC, - Improve the performance in - Increase memory Gaddour
decision based on the fuzzy ETX, battery level energy, and the terms of end-to-end delay, consumption due to the et al. (2014)
logic system parameters. end-to-end delay into one quality network lifetime, and packet implantation of the fuzzy
value. loss ratio. system.
- Compatible with the native - No mobility support.
RPL.
FUZZY OF 2015 - Improve QoS and energy -Use the FIS to combine ETX, delay, -Improve the packet loss ratio - Increase memory Kamgueu
depletion based on the fuzzy and the node’s remaining energy and routing stability consumption due to the et al. (2015)
parameters. metrics into one unique value - Reduce energy depletion. implementation of the fuzzy
(quality). - Shorten end-to-end delay. system.
- No mobility support.
SCAOF 2015 - Provide a specific - Merge energy, reliability, - Enlarge network lifetime. - Incompatible with RPL Chen et al.,
implementation of RPL for A- robustness, and resource contexts - Improve QoS. standard. 2015
LLNs. into a composite using the - Mitigate system costs. - Complex mechanism.
amalgamated routing metrics - Enhance the availability of
approach. the network.
PAOF 2015 - Achieve load-balance in the - Use both ETX and parents count as - Improve load balancing. - No information about energy Gozuacik and
network. routing metrics to calculate the - Prolong the network and mobile nodes. Oktug (2015)
optimal path for routing. lifetime.
- Behave tolerantly in case of
congestion.
DQCA-OF 2018 - Fulfill the requirements of - Use the fuzzy logic system based on - High reliability and low - Increase memory Araújo et al.
specific applications. ETX, HC, and the consumed energy delay. consumption due to the (2018)
routing metrics to select the routes. - Increase the QoS and implantation of the fuzzy
network lifetime. system.
- Reduce energy depletion. - Each application requires a
specific OF. - No mobility
support.
- The change in the number of
sensor nodes and RFID tags is
not considered.
OF-ECF 2019 - Provide reliability and - Employ three routing metrics, - Refine reliability of the - No mobility support. Lamaazi
efficient energy according to which are ETX, energy consumption, network, energy - Inconvenient for highly et al. (2019)
the application requirements. and forwarding delay. consumption, and network dense networks.
- Combine them to present a new stability. - Increase the memory usage.
composite additive metric. - Boost the convergence time
- This metric relies on the weight of and network lifetime
each metric based on the application
requirements.
LA-OF 2020 - Provide accurate link - Utilize machine learning - Improve packet delivery - No mobility support. Saleem et al.
estimation in a dynamic and algorithm, which is learning ratio and energy - Use a single metric, which is (2020)
lossy environment. automata. consumption. not suitable for the diversity
- Each node learns the environment - Better the end-to-end delay of applications in the IoT.
and adjusts the ETX value and control overheads.
accordingly.

34
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

parents’ rank with their respective estimated quality. The obtained re­ 7.4. Mobility enhancements
sults show that the proposed technique has enhanced the network per­
formance in terms of end-to-end delay, packet loss ratio, routing The authors in (Hong and Choi, 2011) suggested using DMR in MSN.
stability, and energy depletion. In (Chen et al., 2015), the authors pro­ This proposal aims to improve the reliability of data transmission. It
posed an OF for a specific application of RPL that would be applied for utilizes rank and LQL information. DMR can effectively construct the
environmental monitoring of A-LLNs. SCAOF is the name of this method. DODAG and provide alternative routes. Besides, DMR uses two mecha­
According to the multipart routing metrics method, it includes nisms for route failures: first, the local repair mechanism, which can
energy-aware, robustness-aware, resource-aware, and reliability-aware swiftly discover multiple alternative routes once all the parent nodes
contexts. By combining the residual energy of A-LLN sensor nodes within the DODAG are unreachable by other nodes by exploiting
with other factors such as RDC, node robustness, and available infor­ neighbor node information. The second one is the global repair mech­
mation in the route cost computation, SCAOF can eliminate the hotspot anism, which can verify the network’s recovery even when neither
issue, lengthen the network’s lifetime, and improve the QoS of A-LLN. neighbor nor parent nodes are reachable. This mechanism reconstructs
The results of the experiments show that SCAOF enables the efficient use the DODAG. Simulation results have shown that the suggested scheme
of memory in resource-constrained devices in A-LLN scenarios, lowering improves the packet delivery ratio and energy depletion. Furthermore,
system costs. Moreover, SCAOF expands the use of IoT technologies in when compared to existing algorithms, the obtained results provide
agriculture. robustness in mobile networks with minimal control message overhead.
A different study for OFs is presented in (Gozuacik and Oktug, 2015), The proposal’s main flaws are its higher memory consumption and
namely PAOF. It aims to fulfill the load-balanced network. PAOF uses insufficient support for the downward traffic pattern (Oliveira and
both ETX and parents count as routing metrics to calculate the optimal Vazao, 2016).
route for routing. Also, it performs preferred parent selection and rank The authors of (Korbi et al., 2012) introduced ME-RPL. This protocol
computing. The acquired results indicate that the proposal outperforms identifies three issues associated with mobility: the first one is that RPL
other routing protocols by achieving a better parent load density, does not distinguish between mobile nodes and fixed nodes. Therefore,
end-to-end delay, and parents change. In addition, PAOF guarantees the there is no route optimization in the existence of mobile nodes. The
load balancing of the network. Therefore, the network lifetime is pro­ second issue is that the trickle algorithm, which is applied in the
longed. Furthermore, it behaves tolerantly in congestion scenarios. In inherent design of static networks to minimize control traffic, performs
(Araújo et al., 2018), DQCA-OF is proposed. The details of this approach slower than desired when the topology changes. The third issue is that
were mentioned in the previous subsection (energy consumption en­ when a node detects changes, the trickle algorithm will reset the timers
hancements). RPL uses OF to construct the DODAG and select the locally to increase the response time until the topology is stabilized
preferred parent depending on link and node metrics. These metrics may again (Korbi et al., 2012). The ME-RPL employs the following mecha­
be single or a combination of them. However, in terms of network per­ nisms, including the modification of the DIO message format to involve
formance, a single metric proved insufficient. Consequently, the the mobility status of the node and changing the preferred parent se­
research envisaged a combination of metrics to improve network per­ lection technique by favoring the selection of static nodes as preferred
formance. This stimulated the authors in (Lamaazi et al., 2019) to put parents rather than mobile nodes. Finally, a dynamic DIS message
forward a flexible OF that relies on three metrics, which are ETX, for­ management mechanism to permit a swift update of the DODAG infor­
warding delay, and energy consumption, called OF-ECF. They created a mation. The obtained findings have shown that ME-RPL improves the
composite additive metric by multiplying each metric by its assigned packet delivery ratio as well as maintains the network’s stability. In
weight. The application’s requirements determine the weight of the (Cobarzan et al., 2014b), the authors introduced an upgraded version of
aforementioned metric. This new OF demonstrated its efficiency from RPL, denoted MM-RPL. It uses ETX as a routing metric. MM-RPL updates
the standpoint of energy consumption, network lifetime, network sta­ the trickle algorithm to permit mobile nodes to move seamlessly within
bility, packet delivery ratio, and convergence time. OF-ECF has several the DODAG. Moreover, it reduces the communications overhead. The
weaknesses, including the inability to support mobility and the proposed approach adopts a reverse trickle algorithm, which lets mobile
impractical nature of high-density scale networks. The emergence of nodes swiftly detect the unreachability of their existing parents to
sustainable cities for diverse applications has significant impacts on the trigger a re-attachment to the DODAG. This timer starts at the maximum
energy consumption of devices. The IoT network’s dynamic and lossy configured value and subsequently decreases after each DIO message
environment presents various challenges, such as wireless communica­ trigger. After reaching the minimum configured value, it restarts the
tion issues, that must be handled in the routing protocol’s architecture. reverse trickle timer if the mobile node is still present. Experimental
In RPL, this issue affects the process of link estimation, e.g., the ETX results highlight that the new scheme permits the mobile nodes in the
metric. That is why the authors in (Saleem et al., 2020) benefited from network to decrease the disconnection times during parent altering.
artificial intelligence, for instance, machine learning algorithms for Furthermore, MMRPL maintains improvement in terms of the packet
adapting the measure of accurate values of ETX in a dynamic delivery ratio and the packet loss ratio. Moreover, the control messages’
environment. overhead is diminished.
They used the learning automata algorithm (Narendra and Tha­ The authors, in (Aijaz and Aghvami, 2015b), suggested Co-RPL as a
thachar, 1974) along with the new OF (LA-OF) to regulate the ETX based new version of RPL. It is built for cognitive networks and aims to support
on the change in the environment. Each node in the network has an LA mobility in RPL. Also, it improves the QoS in terms of end-to-end delay
to assess the LQ. It learns the environment and adjusts the accurate ETX and reliability. The rank is computed by Co-RPL, utilizing ETX as the
accordingly. As a result, it selects the preferred parent or updates it. default metric. An opportunistic forwarding model is used by Co-RPL,
There are two stages of learning: offline and online. The findings of which comprises two main phases: (i) organizing a design to make
LA-OF show that it improves the network performance and outstrips its sure that only the optimal receiver of a given packet will transmit that
counterparts in terms of control overheads, packet delivery ratio, packet (i.e., unique forwarder selection); and (ii) keeping a forwarder set
end-to-end delay, and energy consumption. This protocol can be modi­ where the forwarding node (i.e., next-hop) is chosen opportunistically.
fied to use combined metrics rather than a single metric (i.e., ETX) The suggested model follows RPL’s procedure in DODAG’s construction.
(Darabkh et al., 2021b). The key features of the diversification of OFs for When the root node detects an idle channel, it sends out a DIO message.
RPL enhancements are shown in Table 10. In the option field of the DIO message, each node transfers additional
information within the DIO forwarding process. During the DIO message

35
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

propagation, each node refreshes the neighborhood information. Based results have shown that D-RPL improves energy consumption and the
on this information, each node dynamically selects its neighbors to build end-to-end delay. Also, the packet delivery ratio increases. Besides, it
the forwarder list. The acquired outcomes have shown that Co-RPL di­ decreases the packet’s overhead. In addition, the OF metrics enhance the
minishes energy consumption. The downside of this approach is that it performance of D-RPL, making it more efficient in highly dynamic
has a higher control overhead. The authors in (Fotouhi et al., 2015) networks.
proposed mRPL, which handles the changes in network topology by BRPL: an extension of RPL, was introduced in (Tahir et al., 2018). It
using the default RPL standard, the trickle, and ND algorithms. It aims to integrates RPL with the principles of backpressure routing thus, it sup­
achieve quick and reliable mobility support. The suggested protocol uses ports adapting to several throughputs while considering mobility. BRPL
a proactive fashion. The “smart-HOP” hand-off mechanism has been has two adaptive online algorithms: (i) Quick-Beta: The implementation
employed for mobile nodes by controlling the schedules of the control specifics are fed by Quick-Beta to the mobility-awareness indicator. (ii)
messages (DIS, DIO, and DAO messages) within the trickle algorithm. A Quick-Theta (θ) is an online algorithm that actively modifies BRPL pa­
mobile node selects the new preferred parent based on a specific value. rameters in response to current network dynamics. When the value of θ
To prevent unnecessary routing loops, surrounding nodes in the mobile is close to one, BRPL utilizes only the of the RPL for the packet trans­
node’s child set are omitted. mRPL provides three timers (MDT, CT, and mission. Otherwise, when the value of θ is equal to zero, BRPL employs
RT) to enhance hand-off efficiency by minimizing hand-off delays as backpressure routing principles to perform data transmission. The ob­
well as mitigating network congestion. The simulation results highlight tained results show that BRPL improves network throughput and packet
that mRPl improves the network performance in terms of the packet loss loss ratio. Also, it provides reliability and adaptability for dynamic
ratio, the packet delivery ratio, and end-to-end delay. Moreover, it networks. The main drawback of BRPL is that it significantly increases
achieves higher reliability. end-to-end delay. Other studies introduced a routing solution for a
Another enhancement for mobility in RPL has been implemented for mobile node. In (Bouaziz et al., 2019a), EKF-MRPL was proposed to
healthcare and medical applications. It is named Mod-RPL (Gara et al., support mobility based on the RPL standard and Extended Kalman Filter.
2015). It evaluates RPL for hybrid networks that consist of both mobile The suggested protocol aimed to minimize the signaling overhead and
and static nodes to support the needed short mobility for sensors. The energy consumption based on permitting data routing among
mod-RPL protocol preserves the RPL design in the transmission of con­ non-linearly moving nodes while keeping in mind the limited resources
trol messages, rank calculation, and preferred parent selection. In this of the network. According to EKF-MRPL, mobile nodes only rely on the
approach, preventing mobile nodes from sending DIO messages leads preferred parent to send and receive data without performing any
them to act as leaf nodes. Thus, mobile nodes can join the DODAG but routing processes. EKF-MRPL reduces the number of association changes
cannot act as parent nodes or take part in the construction of the by selecting a new attachment based on the predicted direction. The
DODAG. Consequently, network stability is optimized. The results of the point of attachment to be selected is the one that can operate longer. The
experiments show that Mod-RPL improves control message overhead, observed experimental results reveal that this model enhances position
energy consumption, and end-to-end delay. Besides, it enhances the accuracy, minimizes the cost of signaling, and lowers energy con­
packet delivery ratio. The protocol’s weaknesses include the assumption sumption. Furthermore, it increases the packet delivery ratio. The
that there is always a static node within the range of any other node. In downside of EKF-MRPL is that the end-to-end delay rises.
addition, it does not “add-on” anything to the RPL standard. The The authors in (Bouaziz et al., 2019b) presented a new proactive
mobility issue is addressed in mRPL (Fotouhi et al., 2015) by imple­ approach called EMA-RPL. It is an extended version of the native RPL. It
menting a proactive hand-off mechanism. It is indeed possible that the provided an appropriate platform for real-time applications. Further­
best path will not be selected in the standard RPL. Thus, in mRPL, when more, it is designed for both static and mobile nodes. The proposed al­
it comes to determining the next hop, this algorithm employs a greedy gorithm is capable of maintaining connectivity as well as saving energy
approach. To improve RPL mobility, the authors (Anand and Tahiliani, for mobile nodes. Besides, EMA-RPL integrated three main strategies: (i)
2016) presented mRPL++, a “Smarter-HOP” version of mRPL. Its a more accurate mobility detection procedure that regulates the distance
objective is to evaluate which route is best in mobile scenarios. Of between mobile nodes and their attachments; (ii) a new point of
course, OF is added within that parent selection process to make sure attachment through predicting the novel attachments before discon­
that nodes are informed of routing metrics. The protocol’s disadvantages nection occurs. The prediction process relies on the new location of the
are its reliance on RSSI and the fact that it still suffers from the short­ mobile node, and (iii) an effective substitute method for conserving the
comings of mRPL. mobile node’s energy. The experimental outcomes have shown that
In (Barcelo et al., 2016), the authors introduced a new routing EMA-RPL outperforms existing RPL’s mobility algorithms in terms of
approach for WSNs with hybrid (static and mobile) nodes called KP-RPL. communication overhead, end-to-end delay, and packet loss ratio.
This strategy aims to provide a robust and reliable routing approach. In Furthermore, the suggested design preserves energy and increases the
addition, KP-RPL employs a positioning algorithm based on the forma­ packet delivery ratio. In (Wang and Chalhoub, 2019), a new novel
tion of confidence regions and Kalman filtering. Moreover, a blacklisting mobility and multi-sink improvement algorithm was introduced, enti­
mechanism is exploited to evade disconnections due to positioning tled MRRD+, based on the design of native RPL. Furthermore, it can be
failures. According to the authors’ observations, the suggested protocol used in mobile networks with multiple data gathering nodes. The sug­
has two modes of communication: (i) the anchor-to-anchor (two static gested strategy updates the rank of the nodes according to RSSI values to
nodes), and (ii) the mobile-to-anchor. The first mode handles the native avoid emerging loops. Accordingly, the nodes can smoothly discover the
RPL. But, the second mode is processed by using the Kalman filter and route to reach the root with the best LQ and manage, dynamically, the
blacklisting. The simulation findings demonstrate that KP-RPL improves interval of control messages. Another significant usage of RSSI is the
network reliability and robustness, lowers network infrastructure costs, prediction of nodes’ movements, which is based on the change in RSSI
and prolongs network lifetime for both mobile and static nodes. More­ values. A node detects if one of its preferred parents moves away; it will
over, KP-RPL can improve the ETX with a smaller area. On the contrary, foresee a link failure and try to use another node to be the preferred
the packet delivery ratio increases with a larger region. In (Kharrufa parent. In mobile scenarios, this assists the nodes in updating their
et al., 2017), the authors introduced D-RPL. In this approach, the use of parent selection. Furthermore, the suggested algorithm aids in control­
RPL is examined in dynamic mobility networks for different applica­ ling the total number of managed messages at a reasonable cost. Simu­
tions. The suggested approach is implemented with a novel OF, named lation findings proved that the MRRD + design improves the network
D-OF. Additionally, D-RPL is controlled by a reverse-trickle timer to performance in terms of packet delivery ratio, packet loss ratio,
consult the OF when a node movement occurs, or an inconsistency is end-to-end delay, and the number of transmissions and retransmissions
detected to assist nodes in making an informed decision. The simulation communication overhead.

36
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 11
Summary of RPL enhancements for the mobility challenge.
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

DMR 2011 - Achieve the reliability of - Use rank and LQL information metrics to - Improve energy depletion and - Increase memory usage. Hong and
data transmission in MSN. construct the DODAG. data delivery ratio. - It does not support a Choi (2011)
- Offer alternative routes. - Reduce control overhead. downward route
- Adopt local and global repair strategy.
mechanisms.
ME-RPL 2012 - Improve the mobility - Include mobility status in DIO. - Enhance the packet delivery - Incompatible with the Korbi et al.
support of RPL. - Modify the selection of the preferred ratio and routing stability. native RPL. (2012)
parent.
MM-RPL 2014 - Improve the mobility - Adopt reverse trickle timer design for - Lessen disconnection time. - No mobility detection Cobarzan
support of RPL. mobile nodes. - Optimize the packet delivery approach. et al.
ratio and packet loss ratio. - No information about (2014b)
- Minify control overhead. the nodes’ energy.
- Needs different settings
for mobile nodes.
Co-RPL 2015 - Enhance RPL mobility - Adopt a corona mechanism. - Offer an alternative mechanism - Increase control Aijaz and
support. - Use ETX as a routing metric. to route recovery. overhead. Aghvami
- Provide an opportunistic forwarding - Shorten end-to-end delay. - Modify the format of the (2015b)
approach. - Reduce the packet loss ratio. default RPL messages.
- Relieve energy consumption.
mRPL 2015 - Provide fast mobility - Merge the RPL standard with the smart- - Introduce a mechanism for - Increase control Fotouhi et al.
with higher reliability. hop mechanism. collision and loop avoidance. overhead. (2015)
- Provide three timers: MDT, CT, and RT to - Improve the packet delivery - Choose sub-optimized
monitor the link. ratio. routes.
- Diminish packet loss rate and
delay.
Mod-RPL 2015 - Exploit RPL to provide - Use the same procedure as in the RPL - Reduce the use of control - Oblige the mobile nodes Gara et al.
mobility in hybrid standard. messages. to operate only as leaf (2015)
networks in medical and - Oblige mobile nodes to operate as leaf - Improve the stability of the nodes.
health applications. nodes. network. - No enhancement in RPL
- Reduce energy depletion. design.
- Enhance end-to-end delay and - Limited mobility
packet delivery ratio. management.
m-RPL++ 2016 - Improve the performance - Use hand-off algorithm. - Enhance the packet delivery - Suffer from the Anand and
of mobility management. - Adopt a smarter-hop mechanism that ratio and network lifetime. downsides of mRPL. Tahiliani
takes into account not only RSSI while - Lessen end-to-end delay and - The OF is always (2016)
choosing the next best hop, but also the OF control overhead. dependent on RSSI.
in mRPL. - More simulation
experiments are needed
to test various mobility
topologies.
KP-RPL 2016 - Improve the mobility - Use the Kalman filter and blacklisting. - Use localization techniques. - Susceptible to Barcelo et al.
support of RPL. - Use a positioning algorithm. - Enhance the packet delivery inaccurate positioning. (2016)
ratio. - High energy
consumption.
D-RPL 2017 - Enhance the functions of - Adaptive timer and adaptive DIS - Improve packet delivery ratio, - Limited performance Kharrufa
RPL in dynamic mobile message. energy. consumption, and end- improvements in low et al. (2017)
environments. - Use D-OF as a new routing metric. to-end delay. mobility scenarios.
- Relieve packets overhead.
BRPL 2018 - Provide support mobility - Integrate RPL with the principles of - Improve network throughput - Increase end-to-end Tahir et al.
in dynamic networks. backpressure routing optimization. and packet loss ratio. delay. (2018)
- Offer two adaptive online algorithms - Enhance reliability and - High memory usage.
(Quick-Theta and Quick-Beta). adaptability for dynamic
networks.
- Compatible with the RPL
standard, and it is tested in a
practical, real-time
environment.
EKF-RPL 2019 - Enhance RPL’s mobility - Propose a novel mathematical - Improve the position accuracy. - Lengthen end-to-end Bouaziz et al.
support. framework based on the EKF to detect - Enhance the packet delivery delay. (2019a)
non-linear route movement while taking ratio. - Incompatible with the
direction change into consideration based - Minimize signaling overhead. highly dense network.
on RSSI measurements. - Alleviate energy consumption.
- Proposing a novel method for selecting a - Ensure continuous node
suitable new associate node (point of connectivity.
attachment). - Lower switching between
various attachment points.
EMA-RPL 2019 - Enhance RPL mobility - Integrate three main strategies: an -Shorten end-to-end delay. - Modify RPL control Bouaziz et al.
support. enhanced mobility detection procedure, - Reduce the packet loss ratio message structure by (2019b)
- Preserve the energy. predicting the new attachments before and mitigate communication adding new fields.
disconnection, and an efficient overhead.
replacement method. - Increase the packet delivery
ratio.
- Diminish energy consumption.
(continued on next page)

37
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 11 (continued )
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

MRRD+ 2019 - Improve mobility with - Update the rank of the nodes according - Minimize the overhead in the - Support multiple roots Wang and
multiple root nodes within to the change in RSSI values. network. in one DODAG only. Chalhoub
a single DODAG. - Detect link failure upon the movement of - Improve packet delivery ratio. - It does not use asleep (2019)
nodes. - Decrease packet loss, mode.
- Use dynamic management of control transmission, retransmission, - High energy
messages. and end-to-end delay. consumption.
- Update next-hop neighbors
quickly.
MARPL 2020 - Improve the network - Suggest node’s variability neighborhood - Enhance the network - Increase the DODAG Kniess and
performance in mobility metric based on RSSI. performance in terms of energy reconnection delay. Marques
scenarios. - Design a cross-layer protocol. consumption, packet delivery - the control overhead (2020)
- Suggest a preferred parent unavailability ratio, and control overhead. will increase if the
detection mechanism. - Upgrade the link disconnection disconnection frequently
- Modify the trickle algorithm. prevention procedure. happens.
- Packet loss affects the
packet delivery ratio.

The authors in (Kniess and Marques, 2020) suggested a new increased, and load can be fairly distributed across 6LoWPAN.
enhanced RPL protocol, namely MARPL, for mobile nodes. They employ Furthermore, in comparison with the RPL standard, the overhead of
a cross-layer design. MARPL measures the node’s variability neighbor­ control messages is reduced. Another study in load balancing is the
hood metric by taking the RSSI estimations from the MAC layer (IEEE QU-RPL scheme (Kim et al., 2015a) (Kim et al., 2016). It achieves
802.15.4) and then letting the network layers exploit them. Further­ workload balance by authorizing each node to choose its parent node
more, the proposed protocol is reactive, which means that if one node corresponding to the queue utilization of its neighbor nodes, besides
departs the parent set, the disconnection mechanism will kick in. This their next-hop distances to the border router. This approach successfully
procedure refers to a preferred parent unavailability detection. Also, it achieves load balancing and improves performance in congested net­
modifies the trickle algorithm to regulate the control message trans­ works. The experimental values reveal that QU-RPL is effective at
missions. According to experimental simulations, MARPL optimized reducing queue loss and increasing packet delivery ratios. Additionally,
network performance in terms of energy consumption, packet delivery QU-RPL reduces the end-to-end delay. In (Tang et al., 2016), the authors
ratio, and control overhead. Additionally, it upgrades the link discon­ introduced the CA-RPL protocol for mitigating congestion in LLNs. It
nection prevention procedure. On the other hand, MARPL increases the uses multipath routing and aims precisely to transmit data to the root
DODAG reconnection delay and packet loss, thereby decreasing the quickly and reliably. Enhancing the performance of this protocol is the
packet delivery ratio. This issue needs to be scrutinized. Table 11 pro­ major objective of this work, especially in some applications where
vides a summary of enhanced routing solutions for the mobility emergencies arise, such as monitoring sudden events applications that
challenge. require the avoidance of disconnections. The authors also introduced the
minimal delay metric (or DELAY ROOT), which is calculated using the
Contiki-MAC duty cycling mechanism. Specifically, a node exploits
7.5. Load balancing enhancements merged routing metrics composed of ETX, its rank, the number of
received packets in a period of time, and the DELAY_ROOT to calculate
Load balancing approaches are used to improve the operation of RPL. the cost of a path. When each parent is ranked according to its cost, each
In (Kim et al., 2015a), the authors observed that the networks are sub­ node chooses the two parents with the highest costs. The transmission of
jected to an unbalanced workload, resulting in a shortening network packets over the parents happens proportionally with the path cost.
lifetime. The authors proposed LB-RPL, which aims to balance the Interestingly, CA-RPL has outdone the basic RPL in many terms, such as
workload on LLNs. This is accomplished by detecting workload imbal­ packet delivery ratio, network throughput, packet loss rate, and average
ance and considering how to improve data transmission while keeping latency. That being said, because the amount of metric data carried in a
both link-layer communication features and workload distribution in DIO message is quite sizeable, control message overhead must be
mind. Two approaches are adopted by LB-RPL. The first approach is considered.
forwarding data by exploiting multiple parent nodes. The second Load balancing and mobility are two concerns that affect the per­
approach is based on configuring a timer for DIO message transmission, formance of LLNs. Therefore, LBER-RPL is suggested by the authors in
which is proportional to the workload measured using a counter for the (Sneha and Prasad, 2017) as a new approach to resolve the load
buffer. If an intense workload was on a parent in this round, then more balancing issue and achieve an efficient routing in a dynamic environ­
delays for DIO messages will be added in the next round. The long delay ment in the case of mobile nodes. They employed hand-off scheduling,
reduces the probability that the intensively loaded parent would be in which used multiple timers. Moreover, they utilized RSSI and ETX as
the top k list in its children’s parent table. Thus, the workload imbalance routing metrics for each node to select the preferred parent. Also, to deal
can be relieved. The obtained results prove that the suggested protocol with the load balancing, LBER-RPL detects the preferred parent by
avoids congestion by balancing the workload distribution, maximizes computing the total number of child nodes connected to each neighbor
the packet delivery ratio, mitigates energy consumption, increases the node. The simulation results exhibited that LBER-RPL surpasses other
throughput, and shortens the end-to-end delay. LB-RPL has two weak­ routing protocols in terms of packet delivery ratio and hand-off delay,
nesses that make it incompatible with the RPL standard. The first is that while the control overhead is increased due to the hand-off and
LB-RPL creates new control messages that require modifying the RPL load-balancing procedures. CL-RPL has been introduced for addressing
standard. The second is that the DODAG construction process is the load balancing issue (Taghizadeh et al., 2018). In particular, the
modified. authors proposed a new OF, which is named “context-aware OF”, that
MLEq was proposed by the authors in (Ha et al., 2014). It is a method combines ETX, the parent’s rank, and the usage of the residual energy of
for distributing workload that is both dynamic and balanced. They the parents’ chain in the path towards the DODAG root. Additionally, to
attempted to achieve global load fairness by using water flow behavior make the network more balanced, it suggested the context-aware
as a guide. The simulation results show that throughput can be

38
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 12
Summary table for RPL enhancements for load-balance challenge.
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

LB-RPL 2013 - Achieve load- - Use multiple parents. - Increase the packet delivery ratio. -No mobility support. Liu et al. (2013b)
balancing. - Detecting workload - Shorten energy consumption. - Incompatible with the
- Disseminate the load imbalance. -Improve the throughput and end- RPL standard.
on different routers - Optimize data transmission. to-end delay.
through multiple - Avoid congestion by
routers. balancing the workload
distribution.
MLEq-RPL 2014 - Achieve global load - Use multiple parents. - Optimize throughput and energy -No mobility support. Ha et al. (2014)
fairness, motivated by efficiency. - Incompatible with the
water flow behavior. - Achieve load balancing. RPL standard.
- Reduce message control overhead.
PAOF 2015 - Achieve load - Use both ETX and parents - Improve load balancing. - There is no information Gozuacik and Oktug
balancing in the count as routing metrics to - Prolong the network lifetime. about energy and mobile (2015)
network. calculate the optimal path for - Act tolerantly in case of nodes.
routing. congestion.
QU-RPL 2015 - Improve network - Allow nodes to choose their - Achieve load balancing. - No mobility support. (Kim et al., 2015a) (Kim
performance in parent node based on the - Improve packet delivery ratio and - It cannot be used in et al., 2016)
congested networks. queue utilization of neighbor energy efficiency in congested uncongested routes.
- Achieve load nodes and distances to the routes.
balancing. root. - Reduce the queue loss ratio and
- Disseminate congestion end-to-end delay.
information in the DIO
message.
SSCF 2016 - Aim to increase the - Sink-to-sink coordination - Higher throughput. - No practical experiments. Khan et al. (2016)
network’s throughput framework. - Low energy consumption. - No mobility support.
and lifetime. - Use multiple sinks. -Increase the network lifetime. - Incompatible with the
- Improves the packet delivery native RPL.
ratio.
CA-RPL 2016 - Migrate congestion - Use multipath routing for - Improve throughput, packet - No mobility support. Tang et al. (2016)
by using multiple data transmission based on a delivery ratio, and the end-to-end - Increase in energy
routes. new minimized delay metric. delay. consumption.
- Provide load - Enhance packet loss rate and
balancing. average latency.
- Achieve load balancing.
LBER-RPL 2017 - Fulfill load balancing - Use ETX and average RSSI - Strengthen the packet delivery - Increase control overhead Sneha and Prasad
and efficient routing metrics. ratio and reduce the hand-off delay. due to the load balancing (2017)
in a dynamic - Employ hand-off scheduling. and hand-off procedures.
environment. - Compute the total number of
child nodes connected to each
neighbor node to select the
preferred parent.
CL-RPL 2018 - Overcome the - Consider the status of a - Improve network lifetime. - Slightly higher memory Taghizadeh et al.
problems of RPL under parent chain. - Decrease packet loss ratio. usage. (2018)
heavy and dynamic - Consider the remaining - Shorten energy consumption. - Raise the end-to-end
load networks. queue, energy level of delay.
candidate parents.
- Use ETX as a routing metric.
LoBaPS 2019 - Achieve load- - Use Wake Up Radio and the - Make the network more resilient. - Increase the energy Sampayo et al. (2019)
balancing and always-on feature. - Increase packet delivery ratio and consumption when the
efficient energy. - When the node starts network lifetime. nodes are in the listen
transmitting packet, it turns its - Reduce control overhead and mode and their status is
status on. latency. always-on.
DCRL-RPL 2020 - Improve network - Split the network area into a - Improve network performance in - High memory usage Kumar and
performance metrics diverse set of dissimilar grids. terms of network longevity, end-to- - There is no mobility HariharanDCRL-RPL
by achieving load - Implement the grid head end delay, control overhead, packet support. (2020)
balancing. node selection process to delivery ratio, packet loss ratio, and
choose the head benefiting load balancing.
from the random walk ranking.
- Choose the optimal head
based on the load influence
index, remaining energy, and
the distance from the root
node.
- Design two OFs; the first one
is the O2F for sensitive data
transmission. The other is MOF
for non-sensitive data
transmission.
- Use an adaptive trickle
algorithm.
EL-RPL 2021 - Provide an efficient - Suggest a new OF, which - Reduce the control overhead and - No mobility support. Idrees and Witwit
energy and load amalgamates the highest energy consumption. (2021)
balancing approach. residual energy and the total - Increases the network lifetime.
(continued on next page)

39
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 12 (continued )
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

number of received packets by

the parent.
- Avert the nodes that have
lower ranks from sending DIO
messages.

routing metric, which counts the queue utilization and residual energy reviewed. Table 12 presents a summary of the already stated routing
of the parents’ chain rather than a single parent. The node determines protocols for the load-balance challenge.
the preferred parent and, as a result, the optimal route to the root based
on the new metric. CL-RPL improves the network lifetime and reduces
the packet loss ratio. 7.6. Congestion detection and control enhancements
Wake-up radios are an emerging technology that is intended to
provide efficient energy while being devoid of influencing the reliability Congestion can affect network performance, and thus several pro­
and latency of the network. This technology motivates the authors in posals for CC algorithms for LLNs have been introduced. The authors in
(Sampayo et al., 2019) to integrate it with RPL to develop their sug­ (Michopoulos et al., 2012) have proposed a new CC scheme for DCCC6
gested protocol, namely, LoBaPS. It chooses the parent opportunistically for monitoring traffic in 6LoWPAN networks. The operation of DCCC6 is
by taking advantage of the wake-up radio, always-on feature, and adjusted according to the detection of duty cycling. In this CC scheme,
ultra-low energy consumption. LoBaPS addresses load balancing to routing processes are managed by RPL, and DCCC6 is used to detect
prolong the network lifetime. When the nodes start the packet trans­ jamming by monitoring buffer residence. As a result, when there is
mission, they choose the next hops by waking up all of them in a congestion, the parents will notify each node affected by the congestion,
distributed way. Afterwards, the network can adjust rapidly to shut affecting the rates using the adaptive rate transmission approach.
down the nodes. This makes the network more resilient, which in turn DCCC6 will be in charge of the child notification process. When there is
increases the packet delivery ratio and the network lifetime, besides duty cycling, this procedure will be changed. The obtained outcomes
reducing the latency and control overhead. The shortcoming of this have shown that DCCC6 lowers the packet loss ratio and maximizes the
protocol is the energy consumption increases, in particular, when the throughput. Furthermore, the results indicate that DCCC6 consumes
node is in listen mode and all the next-hops are waked up (always-on little energy and has a short end-to-end delay, as well as a high degree of
feature). The authors in (Kumar and Hariharan, 2020) introduced a fairness.
novel routing protocol to overcome the load balancing challenge. The Griping, Deaf, and Fuse are three distinct cross-layer and completely
new protocol is DCRL-RPL. They split the network area into a diverse set decentralized CC strategies suggested by the authors in (Castellani et al.,
of dissimilar grids. The random walk ranking algorithm is employed in 2014). These strategies use a combination of queue length, buffer length,
the grid head node selection process to choose the head node according and a hybrid of the two. Furthermore, these techniques compare the
to three parameters, which are the distance from the root, load influence parameters versus model backpressure and current UDP-based protocol
index, and remaining energy. To avoid congestion, the head node in stacks. As a result, NS-3 was utilized to determine the results of a
each grid manages its members based on a reputation-based scheduling large-scale simulation experiment. The simulations looked at two sce­
process. For choosing the best parent, DCRL-RPL proposed a novel OF, narios: (i) unidirectional and upwards flows; and (ii) bidirectional
which is named the dual context-based objective function. The term Web-based CoAP flows. In terms of performance, “Fuse” was the best
“context” is classified into sensitive and non-sensitive data transmission option. It detects congestion by combining “Deaf” with clear control
by the Adam deep neural network algorithm. Correspondingly, they messages. This scheme is better for all the measured metrics. Further­
implemented two OFs; the first is O2F for sensitive data transmission, more, it is unquestionably a viable solution for uni- and bi-directional
while the second is MOF for non-sensitive data transmission. traffic.
The MOORA algorithm is employed to choose the preferred parent In (Hellaoui and Koudil, 2015), the authors used a bird flocking
node to send the data message. Finally, the authors developed an model to alleviate congestion in CoAP/RPL/6LoWPAN networks. In this
adaptive trickle algorithm to regulate the control message transmission. intriguing method, the ZoR of a node is the set of its parents and chil­
The simulation results demonstrated that DCRL-RPL improved network dren. The ZoA is the set of direct ascendants of its parents and direct
performance in terms of network lifetime, end-to-end delay, control descendants of its children. Each packet will avoid the congested nodes
overhead, packet delivery ratio, packet loss ratio, and load balancing. in ZoR and follow less congested nodes in ZoA. The filling of a buffer is
This protocol has some flaws, among which are increasing memory taken as a ratio and is used to define the congestion level of a node.
usage and not supporting mobility and security. Passively listening to the data messages is the technique used by the
To address load balancing and efficient energy, the authors in (Idrees authors to understand the buffer filling ratios of other nodes correctly
and Witwit, 2021) suggested EL-RPL. They presented a new parent se­ without having a high overhead. The next hop is chosen based on two
lection procedure depending on a new OF, which amalgamates the complaints: (i) the next hop must be the least congested in ZoR; and (ii)
highest residual energy and the total number of received packets by the the successors of the next hop must be the least congested in ZoA. The
parent. This indicates that EL-RPL chooses the nodes that have the proposed algorithm’s performance was assessed utilizing the COOJA
longest network lifetime and distributes the load amongst the parents. simulator. Nonetheless, the authors demonstrated the impact of shifting
Moreover, they developed the DODAG construction stage by preventing retransmission-related parameters, but they did not present how effec­
the nodes that have lower ranks from sending DIO messages. The tive the bird flocking strategy was at selecting next-hop and mitigating
simulation results show that EL-RPL reduces the control overhead and congestion when compared to existing RPL solutions.
energy consumption. Accordingly, this is reflected negatively on the The authors of (Ma et al., 2016) proposed the GTCC, which relies on
network lifetime. The above-named protocols, PAOF (Gozuacik and changing the parent to mitigate the impact of congestion. GTCC senses
Oktug, 2015) and SSCF (Khan et al., 2016), are other studies of load congestion differently compared to the previous approaches as it con­
balancing approaches. For more details, the subsections, the diversifi­ trols the acquired packet flow rate. This is the packet service rate sub­
cation of OFs and energy consumption enhancements, should be tracted from the packet generation rate. When a positive value is
determined for the net packet flow rate, the congestion probability is

40
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

thought to be high. Once congestion is noticed, a parent node transmits a LQL will determine the node rank. Each node chooses the parent that
DIO message to its children with the congestion notification bit set. The lowers the rank. Consequently, congestion happens when a parent is
DIO message also has the parent information, children’s list, the corre­ picked by an excessive number of nodes. In GTCC, this concern is clas­
sponding transmission rate, and LQL. When receiving DIO messages sified as a potential game called the “parent-selection game.” Simula­
from the parent, a node in the jammed area starts the parent-change tions proved that GTCC reached higher throughput and a lower packet
process, relying on potential game theory. The rank of the parent and loss rate. The limitation of this method is that to compute the potential

Table 13
Summary table for RPL enhancements for the congestion detection and control challenge.
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

DCCC6 2012 - Control and adjust the traffic - Duty cycle aware CC. - Improve both energy - No mobility support. Michopoulos et al.
based on RDC and buffer efficiency and the end-to- - It cannot use uncongested (2012)
occupancy. end delay. routes.
- Minimize throughput.
BPCC 2014 - Migrate congestion in RPL - Use three different cross-layer and - Improve the packet - No mobility support. Castellani et al.
networks. fully decentralized CC schemes. delivery ratio and energy - It does not consider using (2014)
- To alleviate congestion, it uses efficiency. uncongested routes.
queue length and buffer length.
BFCC 2015 - Improve network performance - Detect least congested routes - Improve the packet - No mobility support. Hellaoui and
in congested networks. using a bird flocking technique delivery ratio in the - Increase in energy Koudil (2015)
based on buffer occupancy. presence of congestion. consumption.
- Produce
counterproductive results in
non-congested scenarios.
QU-RPL 2015 - Improve network performance - Allow nodes to choose their - Achieve load balancing. - No mobility support. (Kim et al., 2015a)
in congestion networks. parent node based on the queue - Improve packet delivery - It cannot be used in (Kim et al., 2016)
- Achieve load balancing. utilization of neighbor nodes and ratio and energy uncongested routes.
distances to the root. efficiency in congested
- Disseminate congestion routes.
information in the DIO message. - Reduce the queue loss
ratio, and end-to-end
delay.
GTCC 2016 - Detect the congestion, failures, - Use game theory to discover the - Improve the packet - No mobility support. Ma et al. (2016)
and unavailability of nodes. non-congested routes. delivery ratio and - Increase overheads and
throughput. energy consumption.
CA-RPL 2016 - Migrate congestion by using - Use multipath routing for data - Improve throughput, - No mobility support. Tang et al. (2016)
multiple routes. transmission based on a new packet delivery ratio, and - Increase in energy
- Provide load-balance. minimized delay routing metric. the end-to-end delay. consumption.
- Enhance packet loss rate
and average latency
- Achieve load balancing.
GTCCF 2017 - Awareness of the node and the - Use an adaptive transmission rate. - Enhance the packet - No mobility support. Al-Kashoash et al.
application priorities, energy delivery ratio. - It cannot be used in (2017b)
consumption, and buffer - Maximize the uncongested routes.
occupancy. throughput of the
network.
- Reduce energy
consumption.
- Shorten end-to-end
delay.
- Support node and
application priorities.
OHCA 2017 - Migrate congestion in RPL - Use grey theory. - Improve energy - No mobility support. Al-Kashoash et al.
networks. - Merge three routing metrics, efficiency, buffer dropped (2017c)
including queuing delay, ETX, and packets, and end-to-end
buffer occupancy. delay.
- Transmit packets among non- - Enhance the throughput
congested parent nodes. using an adaptive
transmission rate.
- Support node and
application priorities.
QCOF 2019 - Reduce the congestion. - Employ five metrics. - Boost the packet delivery - No mobility support. Ben Aissa et al.
- Create two control messages ratio and throughput. - Not compatible with the (2019)
(NDR and NDR-Ack). - Diminish energy standard RPL.
- Transmit the data according to the consumption.
node’s priorities.
- Develop multiple DODAG
procedure in the case of route
crashes.
Brad-OF 2020 - Mitigate the congestion and - Integrate three metrics: remaining - Reduce energy - No mobility support. Vaziri and
extend the network lifetime. energy, delay, and ETX metrics. consumption. - Does not investigate in a Haghighat (2020)
- Introduce a new metric Node’s - Extend the network dense network.
Traffic Intensity. lifetime.
- Decrease the packet loss
ratio.

41
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

function, each node must have information on the transmission rates important and effective, has not been examined. Importantly, CA-RPL
among parent candidates and their children. In high-density networks, (Tang et al., 2016) and QU-RPL (Kim et al., 2015a) (Kim et al., 2016)
the information carried in a DIO message increases along with the are two congestion detection and control enhancements that mitigate
number of children. congestion in the networks. These proposals were explained, compre­
Many studies adopt non-collaborative game-theoretic designs to hensively, in the load balancing enhancements subsection. Table 13
solve the congestion problem in RPL-based networks. For example, the summarizes RPL enhancements related to the congestion detection and
authors of (Al-Kashoash et al., 2017b) introduced a simple strategy to control challenge.
control congestion called GTCCF. The IoT application requirements are
fulfilled by this strategy, which is observant of both node and applica­ 7.7. Cross-layer issue and QoS enhancements
tion significance. Additionally, GTCCF is dependent on the unique
characters of IEEE 802.15.4, IPv6, and the 6LoWPAN protocol stack. The According to the abovementioned studies, many techniques for
game is designed with the buffer residence size, energy consumption, multicast have been developed to enhance routing reliability and QoS.
and node and application priority in mind. Experimentally, it was The studies proposed SMRF (Oikonomou et al., 2013), ESMRF (Abdel
indicated that the new model advances the QoS properties in jammed Fadeel and El Sayed, 2015), and BMRF (Gastón Lorente et al., 2017) to
networks as for throughput, end-to-end delay, energy consumption, and better control the multicast messages in RPL. The captured findings have
packet loss ratio. Plus, the existing algorithms are outperformed by shown that the mentioned protocols perform better than the trickle al­
GTCCF in attaining a weighted fairness index. gorithm. Also, they proclaimed that the use of link-layer broadcast and
OHCA was introduced by the authors in (Al-Kashoash et al., 2017c). link-layer unicast improves reliability, although the costs of this reli­
This strategy considers both resource control and traffic control. A DIO ability improvement are relatively high due to energy consumption and
message holding congestion information is sent by a given parent to the time delay.
children if it detects that the arrival rate exceeds the service rate. This The authors in (Ancillotti et al., 2014a) adopted a new approach to
situation is where the congestion condition is met. Once a child node the cross-layer design of RPL. It targets enhancing the reliability of data
receives that DIO message, it operates with the resource control delivery. The proposal was named RPLca+. It is considered an improved
approach to select a non-congested parent. If no legitimate (or link estimation capability, a simple routing optimization, and efficient
non-congested) parent is found, the traffic control approach is applied. management of neighbor tables. These novel techniques let a node in
For resource control, a multi-attribute decision making technique called RPL efficiently discover the quality of individual links and operate a
GRA is used to solve the multi-criteria optimization problem that models more informed next-hop selection. Furthermore, the implementation of
the parent selection process in OHCA. The OF in OHCA uses three this scheme is able to synchronize RPL with the IP neighbor tables to
metrics: (i) buffer occupancy, (ii) ETX, and (iii) queuing delay. As for increase the storage sustainability of information. The synchronization
each parent candidate, the rank is calculated in the GRA procedure process is operated in one direction, in a way that the insertion and
while considering diverse features of the routing metrics. GRA is replacement processes in the RPL table are reverberated in the IP table
considered very simple. This feature is vital in resource-constrained but inverted. At the end of this point, it is worth noticing that the pro­
networks. When the resource control strategy collapses (i.e., no more posed model can be interoperated with the conventional RPL imple­
legit parents are found), the next less congested parent will be chosen. In mentation. Because of this, RPLca + does not modify the fields of the
that instance, the traffic control strategy will be applied, relying on the RPL’s control message. Moreover, this model achieves noticeable im­
node, the application priorities, and the received congestion informa­ provements in the packet delivery ratio.
tion. In OHCA, the node’s transmission rate adaptation is characterized In (Ancillotti et al., 2014b), the authors proposed the LLQE proced­
as a constrained optimization problem that is solved with network utility ure. At the time of DODAG’s creation, it used the trickle algorithm,
maximization. Simulation results reveal that OHCA increases the which was first presented in the standard RPL, to adjust the flooding of
throughput, attains fairness, and shortens end-to-end delay. Plus, energy control messages and also to manage the transmission of broadcast
consumption is relieved, and the packet loss rate is reduced. probe messages. Such probing packets are based on modifying DIO
Another enhancement for congestion-aware protocols based on RPL messages to ensure seamless connectivity of the newly proposed link
is QCOF (Ben Aissa et al., 2019). To eliminate congestion and time estimation technique. This also permits simultaneously evaluating the
feasibility, QCOF integrates five metrics: node rank, energy feasibility, LQ while broadcasting routing information. LLQE has been integrated
link capacity, and total input/output data. QCOF transmits data in with Contiki’s RPL prototype implementation. The observed outcomes
proportion to the node’s priorities. When the route crashes, the nodes illustrate that LLQE is capable of estimating the quality of the links to
along this route request new DODAG roots. In other words, they employ neighbors while maintaining a small overhead, which leads to better
multiple DODAGs procedure. QCOF provides two new RPL control routing decisions. Additionally, LLQE improves the packet delivery ratio
messages, NDR and NDR-Ack, to accomplish this task. An NDR is a and achieves the stability and reliability of the data transmission pro­
multicast message used when the node asks for a new DODAG root, cess. The authors in (Lee et al., 2016) developed a protocol architecture
whereas an NDR-Ack is the reply message to the NDR message. The new that enables RPL to run on top of BLE technology. They divided their
OF selects the best path to the root, whereas each node chooses the work into two parts. Initially, to attain efficient energy and reliable
preferred parent to transmit the data via it. The simulation experiments routing, they utilized the advertising and data channels of BLE to
demonstrate that QCOF improves the packet delivery ratio and determine synergistic impacts between RPL and BLE. After that, they
throughput. On the other hand, it diminishes energy consumption. The created ALBER, an adaption layer between RPL and BLE that integrates
target of the work in (Vaziri and Haghighat, 2020) is to extend the RPL and BLE operations. In more detail, this adaption layer enables the
network lifetime along with alleviating the node’s congestion to prevent following: the broadcast of RPL control messages over BLE, the use of
the intense traffic load on the network. BradOF is the name of a new BLE LQ, which can be used to calculate the routing metric of RPL, and
proposal that combines three metrics: remaining energy, delay, and the update in the routing table associated with BLE connection super­
ETX. In addition, they introduced a new metric called “Node’s Traffic vision. They compared their work with RPL over IEEE 802.15.4. The
Intensity,” which was developed to discover the congestion and results provide evidence that RPL can run on top of BLE with a packet
accordingly avoid it. Simulation results indicate that BradOF improves delivery ratio of 100%. Also, the duty cycle is downgraded.
energy consumption, which in turn extends the network lifetime. The advent of IoT applications necessitates not only the maintenance
Moreover, the packet delivery ratio has decreased. It cannot be missed of durable communication between nodes but also the management of
that the number of nodes is slightly small, which means that the scheduled medium access. Over and above, it uses channel hopping and
congestion considering high-density networks, which are very frequency range. Implementing these techniques needs synchronization

42
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

of all devices in the network. Thus, the authors in (Vera-Pérez et al., adaptively and jointly to achieve higher throughput and more reliable
2018) attempted to find a new beacon advertising mechanism depend­ routing. Owing to the “fixed and uniform” transmission power technique
ing on the trickle algorithm for the MAC TSCH layer and the timers in and construction routing topology in RPL, which is based on HC and LQ
the network and network layers. The new solution combines metrics, a significant amount of bandwidth was lost due to load imbal­
IEEE802.15.4e with its TSCH modes and RPL. It improves connection ance and hidden terminal issues. PC-RPL (Kim et al., 2020), for example,
success, joining time, and energy depletion. TSCH-RPL still needs is a proposed cross-layer protocol that is designed on top of RPL and uses
enhancement. For instance, the variable enhanced beacons period pro­ not only an adaptive and joint control mechanism for transmission
cedure needs to be dynamic rather than static, as it is currently accepting power and routing topology but also an RSSI threshold control value.
just two values. On the other hand, SG is one of the applications that uses PC-RPL suffers from load imbalance and hidden terminal issues. To
RPL. SG is composed of various heterogeneous applications. Each balance the load, PC-RPL uses the RSSI threshold adjustment mecha­
application has a special need that is different from others with regard to nism. Also, to tackle the hidden terminal issues, the node decides to trim
priority, latency, and reliability. The main concerns in SG are data pri­ its parent list and chooses the closest parent by increasing the threshold
oritization and routing. RPL enables the usage of multiple instances, and modifying the transmission power. The results demonstrate that
where each instance has its own OF. Consequently, RPL guarantees PC-RPL decreases the packet loss ratio, increases the bandwidth, and
traffic differentiation at the network layer. Nonetheless, QoS differen­ resolves the load imbalance and hidden terminal issues under
tiation is not authorized by the OF and its related routing metrics. To high-density traffic.
address this problem, the authors of (Nassar et al., 2018) proposed ELITE is an elaborated cross-layer OF to achieve efficient energy in
OFQS, a new OF with a multi-objective metric (i.e., mQFQS), which RPL (Safaei et al., 2021). It suggests a new routing metric named SPR,
exploits the dynamic quality of the communication links, delay, and which stands for the number of transmitted strobes per packet sent due
residual energy. OFQS is scaled according to the number of instances, to RDC guidelines in the MAC layer. More to the point, ELITE utilizes the
which are categorized based on their traffic, resulting in a QoS differ­ acquired information from the RDC protocol in the MAC layer and
entiation that is tailored to the specific needs of a wide range of SG transmits it to the route selection metric in the network layer. Through
applications. The results show that OFQS improves the packet delivery the use of relative phase shifts between the transmitter and receiver
ratio, latency, and energy consumption. Accordingly, it prolongs the nodes, SPR is conjoined with asynchronous MAC protocols. The sug­
network lifetime. OFQS needs to be investigated from the perspective of gested protocol finds the optimal route that inflicts the least number of
the fixed parameters for the chosen instances. These parameters can be strobe disseminations on its nodes. As a result, the node’s energy con­
dynamically adjusted using artificial intelligence techniques such as sumption is reduced. The experiments’ findings revealed that ELITE
machine learning and fuzzy logic to attain the most appropriate values. reduces the energy consumption in each node due to minimizing the
Moreover, the maximum number of instances and the upshot of one number of strobes per packet. What is more, ELITE improves network
instance on another while both are running together are questionable. performance in terms of control overhead, end-to-end delay, and reli­
However, the authors in (Mohammadinejad and Mohammadhoseini, ability. Essentially, other studies regarding QoS enhancements, QoS_RPL
2019) suggested a new enhancement to the reliability of RPL in SG (Belghachi and Feham, 2015), multiELT-RPL (Iova et al., 2015), and
neighborhood area networks. The three main operations in the new Pro-RPL (Khelifi et al., 2015) were illustrated in the energy consumption
protocol are: expanding the cache at each node, letting the node have enhancements subsection. In addition, the subsection on diversification
two preferred parents instead of one, and grouping the foregoing pro­ of enhancements surveyed OF-FL (Gaddour et al., 2014), FUZZYOF
cedures together. The simulation results were carried out on the OPNET (Kamgueu et al., 2015), and SCAOF (Chen et al., 2015). Table 14 pre­
simulator and achieved the target of improving reliability. Conversely, sents a summary of the previous RPL enhancements for cross-layer and
this protocol increases the latency. QoS challenges.
SOF-QoS is an innovative fuzzy logic-based process that integrates
three metrics: RSSI, HC, and NE (Lee et al., 2020). The output is the 7.8. Authors’ insights
preferred parent, which provides the optimal route to the DOAG root. It
is determined based on the relative weights of each metric. This OF As we mentioned before, RPL is a widely used routing protocol for
divides the services into three categories: regular data, video packets, the IoT environment. Having said that, the standard RPL obviously has
and emergency message transmission. This classification needs various some issues concerning reliability, packet delivery ratio, mobility, and
QoS, with varying weights for different metrics for each service or efficient energy usage. This motivated the research to overcome these
application. SOF-QoS determines the importance of metrics based on the limitations. Thus, the previous subsection discussed the most recent and
type of service. The first service is network lifetime-aware, in which potentially best solutions for optimizing RPL. Comparison tables artic­
SOF-QoS provides priority to the energy metric by increasing its weight ulating the most crucial aspects (strengths and limitations) of the
among normal data services. The second service is packet delivery mentioned protocols were included at the end of each subsection. As a
ratio-aware, in which the RSSI metric is given a higher weight. The third result, some key aspects of existing algorithms could be observed. Ac­
service is delay-aware, in which the RSSI and HC metrics are distributed cording to the results reported in each paper, the preponderance of the
equally with the emergency message service. In terms of reliability, investigated solutions can improve the packet delivery ratio and mini­
SOF-QoS outperforms the other compared protocols. mize energy usage. Even though these solutions can boost the standard
As mentioned earlier, OF plays a significant role in selecting the RPL, the resulting solution is generally lengthy and complex, necessi­
preferred parent to choose the optimal path. In light of this, the authors tating the nodes to have more storage and processing capacity.
in (Moradi and Javidan, 2020) put forward T-OF, which uses three Importantly, many applications adapted to RPL have many re­
metrics to balance the energy of nodes and extend the network lifetime. quirements to be solved, including mobility, security, and diversification
Those metrics are ETX, energy consumption, and a new metric named of OFs. The existing solutions seek to handle these issues, but still have
NOCS, which represents the minimum value of the number of children limitations as reported in the comparison tables. For instance, applica­
and siblings. T-OF combines those metrics depending on their weights, tions that require P2P communications have excessive control overhead
which are adjusted dynamically during the network lifetime. The and energy consumption. Other applications impose the use of multicast
simulation results confirm that T-OF balances the energy of nodes, techniques (i.e., SMRF, BMRF, and ESMRF), which improve the RPL in
thereby prolonging the network lifetime. Furthermore, the proposed terms of energy consumption and QoS. However, they flood the network
protocol increases the packet delivery ratio and shortens the with control messages, resulting in increasing control overhead. More to
end-to-end-delay. the point, mobile applications that depend on the nodes’ movements,
The routing topology and transmission power must be controlled require tracking the change in topology as the nodes move. Although

43
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 14
Summary table for RPL enhancements for the cross-layer issue and QoS challenges.
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

SMRF 2013 - Improve multicast data - Design a cross-layer - Achieve significant - Support only downwards Oikonomou and
transmission. approach that enhances RPL improvement in terms of multicast traffic within the Phillips (2012)
- Adopt a cross-layer for multicast transmission. end-to-end delay and DODAG (the absence of upward
approach. - Design a trickle multicast energy efficiency. multicast support).
algorithm. - Reduce packet delivery - Only messages sent from the
ratio. node’s preferred parent are
- It does not require geo- handled, while messages received
location information. from child nodes are not handled
- Low memory or retransmitted as a result.
requirements and code
size.
- Evade duplicated
multicast packets process.
RPLca+ 2014 - Aim to enhance the - Adopt a cross-layer model - Compatible with the - Increase the overhead in the Ancillotti et al. (2014a)
reliability of data delivery. for LQ estimation and conventional RPL. network.
- Improve link estimation efficient management of - Improve the packet - No mobility support.
capabilities. neighbor tables. delivery ratio.
- Permit nodes to discover
individual links’ quality and
to perform a more informed
next-hop selection.
OF-FL 2014 - Offer a configurable - Combine the following - Improve the performance - Increase memory consumption Gaddour et al. (2014)
routing decision based on metrics: HC, ETX, battery in terms of the end-to-end due to the implantation of the
the fuzzy logic system level energy, and the end-to- delay, network lifetime, fuzzy system.
parameters. end delay into one quality and packet loss ratio. - No mobility support.
value. - Compatible with the
native RPL.
LLQE 2014 - Improve LQ estimation in - Exploit the trickle algorithm - Improve the packet - Add extra overhead for the Ancillotti et al. (2014b)
RPL. for LQ estimation. delivery ratio. network.
- Adopt a cross-layer - Compatible with the - Increase in energy consumption.
approach. native RPL. - Lengthen end-to-end delay.
- No mobility support.
QoS_RPL 2015 - Satisfy the demands for - Utilize the remaining - Lower energy - Packet delivery ratio is slightly Belghachi and Feham
efficient energy and QoS in energy and the transmission consumption. decreased. (2015)
LLNs. delay as routing metrics for - Prolong the network - Increase communication
the next-hop selection. lifetime. overhead messages.
- The OF is built using the
ACO method.
multiELT- 2015 - Detect energy-bottleneck - Use ELT as a routing metric. - Control overhead - Change control message Iova et al. (2015)
RPL nodes. - Provide a multipath messages and energy structure.
- Achieve energy approach. consumption are reduced. - No mobility support.
balancing. - Use of a procedure to - Minimize the number of - Raise the usage of memory
-Maximize the lifetime of control the parental transmissions. slightly.
nodes. exchanges. - Reduce the number of
DODAG reconfigurations.
ESMRF 2015 - Allow nodes to be able to - Multicast traffic sources - Support both upward and - Emerge a bottleneck at the root Abdel Fadeel and El
send multicast traffic encapsulate their multicast downward traffic for node. Sayed (2015)
upward and downward. packets and send them to the multicast data. - High end-to-end latency.
root that forwards the - Improve the reliability of - Increase in energy consumption.
multicast packet. the network.
- Improve the packet
delivery ratio.
FUZZY OF 2015 - Improve QoS and energy -Use the FIS to combine ETX, -Improve the packet loss - Increase memory consumption Kamgueu et al. (2015)
depletion based on the delay, and the node’s ratio and routing stability. due to the implementation of the
fuzzy parameters. remaining energy metrics - Reduce energy depletion. fuzzy system.
into one unique value - Shorten end-to-end delay. - No mobility support.
(quality).
SCAOF 2015 - Provide a specific - Combine energy, reliability, - Enlarge network lifetime. - Incompatible with the RPL Chen et al. (2015)
implementation of RPL for robustness, and resource - Improve QoS. standard.
A-LLNs. contexts routing metrics and - Mitigate system costs. - Complex mechanism.
convert them into a - Enhance the availability
composite metric. of the network.
Pro-RPL 2015 - Exploit a proactive - Use a suffering index that - Pro-RPL is not a complex - No mobility support. Khelifi et al. (2015)
approach to establish reflects the cost of network approach. - No practical experiments.
failures and detect them. failures. - It does not consume much
- Predict the problem before memory.
it occurs. - Reduce the packet loss
- Trigger the local repair ratio and energy
mechanism. consumption.
- Enlarge the network
lifetime.
(continued on next page)

44
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Table 14 (continued )
Enhanced Year Aim Strategy Strengths Limitations Ref.
protocol

ALBER 2016 - Attain efficient energy - Utilize advertising and data - Downgrade the duty - Increase end-to-end delay. Lee et al. (2016)
and reliable routing. channels of BLE to determine cycle.
synergistic impacts between - Improve packet delivery
RPL and BLE. ratio.
- Implement an adaption - Attain high energy
layer between RPL and BLE efficiency.
(ALBER), which combines - Provide a reliable multi-
RPL and BLE processes. hop routing model.
- Reduce parent changes.
BMRF 2017 - Enhance SMRF’s - Combine the best features of - Lessen radio - Increase memory usage. Gastón Lorente et al.
multicast data forwarding. RPL multicast and SMRF. transmissions and energy - Increase end-to-end delay. (2017)
- Offer a choice between Link consumption.
Layer broadcast and Link - Higher packet delivery
Layer unicast depending on a ratio.
threshold value. - Improve reliability.
TSCH-RPL 2018 - Maintain durable - Combine IEEE802.15.4e - Improve the connection - Modify the variable Enhanced Vera-Pérez et al. (2018)
communication between with its TSCH modes along success and joining time. Beacons period procedure to
nodes. with RPL. - Reduce energy depletion. adapt dynamically.
- Operate scheduled - Find a new beacon
medium access. advertising mechanism
- Use channel hopping and depending on the trickle
frequency range. algorithm for the MAC TSCH
layer.
OFQS 2018 - Employ multiple - Suggest a new OF (OFQS), - Improve the packet - The maximum number of Nassar et al. (2018)
instances to achieve which adjusts to the number delivery ratio, latency, and instances is questionable.
reliability. of instances according to energy consumption. - Choose the parameters
their traffic. - Prolong the network dynamically instead of fixed
- Propose a new metric, lifetime. values.
mQFQS, that exploits the
dynamic quality of the
communication links, delay,
and residual energy.
ER-RPL 2019 - Improve data - Increment of the cache at - Decrease packet loss. - Increase the latency. Mohammadinejad and
transmission in each node. - Increase packet delivery - Not compatible with standard Mohammadhoseini
neighborhood area - Allowing the node to have ratio. RPL. (2019)
network of SG two preferred parents instead
communication network of one.
using RPL. - Group the aforesaid
procedures together.
SOF-QoS 2020 - Design an OF that meets - Apply fuzzy logic, which - Prolong network lifetime. - Increase memory usage due to Lee et al. (2020)
the requirements of combines three metrics: RSSI, - Shorten end-to-end delay. the complexity of fuzzy logic.
different applications to HC, and NE. - Increase packet delivery - Is not being investigated for
achieve reliability. - Provide each metric with a ratio. high-density networks.
weight that corresponds to
each service.
- Select the preferred parent
based on the output of the
fuzzy logic.
T-OF 2020 - Balance the energy of - Combine ETX, energy - Balance the energy and - Is not examined for high-density Moradi and Javidan
nodes. consumption, and a new protract the network networks. (2020)
- Extend the network metric named NOCS to gain a lifetime. - No support for mobility.
lifetime. new OF named T-OF. - Shorten the end-to-end
- The weights of metrics are delay.
adjusted dynamically. - Improve the packet
delivery ratio.
PC-RPL 2020 - Mitigate the hidden - Use an adaptive and joint - Use real embedded - Does not tackle the wireless Kim et al. (2020)
terminal impacts. control procedure for the devices to implement the interference issue.
- Attain load balancing. transmission power and protocol. - Increase end-to-end delay.
routing topology. - Improve the throughput
- Employ a threshold control and packet loss ratio.
value of RSSI. - Resolve the issues
concerning load imbalance
and hidden terminal.
ELITE 2021 - Design an energy efficient - Elaborate cross-layer OF to - Reduce the energy - Some powerful routing metrics Safaei et al. (2021)
OF. achieve efficient energy. consumption by each node can be used as an OF that require
- Apply a cross-layer design - Introduce SPR routing due to minimizing the less energy consumption and
by considering MAC layer. metric. number of strobes per have less complexity than SBR,
packet. such as energy per packet.
- Strengthen the network
performance regarding
control overhead, end-to-
end delay, and reliability.

45
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Fig. 19. Total number of scholarly publications related to various RPL challenges in the period 2010–2021.

several solutions have been presented in this area, they have many 8. Recommendations and future directions
limitations with respect to end-to-end delay, the size of control message
due to the added fields, and the expansion of routing tables of the nodes The following subsections provide the reader with an overall view of
due to the frequent movement of the nodes. the current status of RPL, its challenges, and the limitations of its up-to-
As for solutions that draw upon cross-layer design, they induce the date enhancements. Based on this, the recommendations and future
need to deploy a specialized MAC protocol as well as an increase in the directions are concluded.
end-to-end delay. In the case of the solutions introduced to achieve load
balancing and reduce congestion, they had accomplished their goals. On
the other hand, end-to-end delay, length of control messages, algorithm 8.1. Research history analysis and recommendations
intricacy, and memory usage of devices are increased. Because each
application has its own unique needs, the diversification of OFs in RPL is The primary motivation for conducting such research is to direct
a major demand for IoT applications. In order to do so, IoT applications researchers to the most recent and future work in the RPL routing pro­
with constraint devices must have low energy consumption, a higher tocol. An extensive body of papers has been published from 2010 to
packet delivery ratio, a shorter end-to-end delay, and less control 2021 to cover the challenges in the RPL in various conferences and
overhead. The OF0 and MRHOF (i.e., RPL’s standard OFs) have serious journals. The standardization of RPL as the routing protocol of the IoT in
flaws. As a result, many of the solutions presented in this context looked 2012 paved the way for many researchers to make all possible efforts
at the differences between utilizing a single metric as an OF and using from different fields and domains, for instance, academics, communi­
multiple metrics to see which is the best option and showed excellent cations, and industries. These efforts are required to tackle the main
results. The preponderance of solutions, in essence, employ FIS to challenges of RPL, improve RPL’s performance, and begin to employ
integrate multiple metrics as the system’s input, with the output rep­ RPL in industrial applications. To give an idea of the progressing
resenting neighbor quality. The FIS is an artificial intelligence technique research work in the enhanced solutions concerning the challenges
that makes decisions based on experience and human knowledge. On the raised by RPL, we used three different databases, i.e., IEEE Xplore,
other hand, their implementation is more complicated and requires Google Scholar, and Scopus, to draw a conclusion about the number of
more computational resources than conventional OFs. What is more, the publications from 2010 to 2021. According to Fig. 19, which displays the
most commonly used simulators (e.g., COOJA) test the theoretical and observed research paper count per year and challenge, RPL has drawn
statistical part for the majority of contemporary RPL-based solutions to more attention, whereas more topics are being covered every year since
evaluate the network performance. Despite the fact that the simulators its standardization. In the first few years (2010–2013), the core of the
provide a wider range of various situations and node densities, the re­ research was on probing RPL operations (i.e., upward and downward
sults may not reflect the approach’s real and actual performance. Un­ modes, multicast, MOP, and P2P) and ameliorating energy efficiency, as
fortunately, only a small number of modern techniques have been tested well, regardless of the missing functionalities. Though, in later years
in the real environment. Although these considerations are not required, (2014–2015), the trend turned towards appending more functionalities
the advent of the smart city paradigm and its applications require the and developing the quintessential design of RPL, such as adding mobility
translation of these solutions into practical ones. Not only to this extent, and security concepts. Over time, precisely from (2016 -till now), mul­
but rather, it is crucial to note that they can point you in the correct tiple issues were covered, such as multipath routing, mobility, QoS, load
direction for a routing solution that works well and satisfies the ambi­ balancing, cross-layer design, and congestion, which witnessed exten­
tions of citizens. sive paper research that generated multiple valuable improvements on
In light of the preceding arguments, we can guide the research and RPL. Not only that, but also the direction shifts to the development of
designers of new RPL-based solutions to address the main limitations of new OFs that satisfy the application requirements, whether using a
prior discussed solutions, such as the sensitivity of IoT nodes in terms of single metric or multiple metrics, according to its prominence. As things
low memory and low power, by finding approaches that alleviate al­ stand now, our recommendations are based on the extensive analysis
gorithm intricacy, increase control message size, and simultaneously and the number of publications for each challenge per year. Therefore,
meet the application requirements while enhancing the network per­ we recommend that researchers move towards utilizing RPL in indus­
formance from all aspects by making a trade-off between them as well. trial areas, designing more applications based on it to be more practical,
as well as adopting cross-layer designs for its enhancements that cover

46
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

different IoT technologies rather than IEEE 802.15.4 only. Additionally, maintain network connectivity (Panichpapiboon et al., 2009). The
taking advantage of the drawbacks of the current few load-balancing second is applying clustering mechanisms to choose the best trans­
approaches for enhancing RPL is highly recommended. Interestingly, mission range for the cluster heads (Jarry et al., 2006) (Song et al.,
to mitigate the congestion over DODAG routes, it is highly recom­ 2009). To the best of our knowledge, there is no RPL research that
mended to employ multiple instances, multiple DODAG roots, machine handles the process of selecting the best transmission power to
learning algorithms, new optimization methods, and clustering mecha­ choose the optimal path in terms of efficient energy and reliability.
nisms. Ultimately, enabling RPL security for preventing RPL from Further solutions may adjust the optimal power level for each node
various attacks is of massive interest, taking into account that this that transmits data. This improves the energy depletion overall of the
challenge is perfectly addressed in this document. network.
• Efforts to minimize resource consumption: Energy consumption is
8.2. Future directions crucial in designing protocols for LLNs. Implicitly or explicitly, the
new routing protocols try to save energy. Memory capacity is also a
In conclusion, in this subsection, we suggest a set of research di­ critical limitation. The matters strictly related to RPL design are
rections and principles to heed in the design of new proposals based on summarized as follows: (i) Interoperation between differently
the native RPL. It is important to point out that these inferred future structured devices is not granted by the current implementations of
directions are the result of considerable efforts made on this survey after the standard protocols because they are affected by resource con­
investigating roughly 60 solutions over RPL and taking their limitations straints. (ii) Because adaptive data rates, interference avoidance, and
and shortcomings into consideration. As a result, the findings in this multi-antenna diversity are supported by advanced radio features,
document will be a valuable resource for network engineers looking to they open up new research opportunities.
improve more effective RPL routing solutions for IoT use cases. • Investigation based on real-world implementation and experi­
ments by utilizing hardware platforms: As indicated earlier, the
• Considering the heterogeneity of nodes: One or two devices’ types emergence of the smart city paradigm and its applications require the
are considered in the existing routing solutions. Yet, the IoT’s main deployment of high-density nodes to ensure efficient routing. How­
concept is based on interoperability across diverse devices. Message ever, the majority of solutions suffer from the absence of real and
processing ratio and, thus, the performance of routing protocols are actual implementations to make them more practical. Thus, the trend
affected by varying hardware settings. Therefore, the diversity of these days is towards practical solutions that serve the interests of all
devices is well-thought-of as a serious factor and must be managed in sections of the country by making effective use of the supported
the development of new routing solutions for the IoT. applications, the country’s prosperity through economic growth,
• Interoperability with standard protocols: Dominant routing pro­ and, eventually, the citizens by improving their quality of life and
tocols are adopted by the majority of the explored routing solutions. increasing their satisfaction. This can be achieved by integrating RPL
Thus, these solutions approximately use all the structures of the with new technologies, which can be used for a wide range of
standard protocols, for example, ad-hoc on-demand distance vector applications.
and RPL. Still, a few of the offered solutions support interoperability • Considerations on multiple RPL instances: Multiple instances in
with the base routing protocol. This gap counts these new solutions an LLN are authorized by RPL to have various QoS requirements.
hard in a currently working IoT network; because it would need a Every instance could have a different OF, and beyond that, they
revive. Thus, it is recommended that future routing solutions would could have dissimilar routing metrics and constraints. According to
be able to collaborate with the standard protocols while ensuring the Kim et al. (2017a), there was a lack of studies on multiple instances
synchronicity of both protocols in the same network. To achieve this from 2010 to 2016. According to our statistical analysis, the count of
prerequisite, the network should avoid significant changes in the multiple instances studies from 2013 till now is 10 papers, which is
messages’ format (fields) and the processing of control messages. quite a few in comparison with other challenges. Although this vision
• Need an adequate trade-off between the network aspects: The will be modified in the event of the appearance of devices with
requirements of the 6LoWPAN and RPL routing protocols can differ additional memory, meaning they will be able to handle the design of
according to the applications, as mentioned before in this work, some multiple RPL instances, it is still using multiple instances is consid­
applications need to achieve efficient energy, while others intend to ered one of the requirements, which makes the design of RPL more
attain reliability. Nevertheless, the requirements of applications are complex and should reconsider the constraints of resources in
frequently synchronized with each other. For instance, the embedded nodes.
enhancement in network load balancing may influence, in a negative • Relationship between TCP and RPL: There is general agreement in
way, the end-to-end delay. As a result, proposed routing solutions the literature on traditional WSNs that the TCP consumes resources
must search for an adequate trade-off between the synchronized wisely. For that reason, TCP is not a satisfactory candidate. Despite
requirements of the application. this, TCP has attracted attention recently because IPv6-based LLNs
• Prolong the network’s lifetime for the dense and large net­ have become diffused. The TCP usage guidelines in IoT (Kim et al.,
works: Within LLNs, the sensor nodes are power-limited devices, and 2015b) confirmed that one of the key motives behind the use of TCP
replacing them once their energy is consumed is considered an in constrained networks is that the use of UDP, in existing in­
extremely challenging mission. Hence, to improve the network per­ frastructures, could be partial or blocked. The LLN-based TCP has
formance, it is important to ensure satisfactory network coverage of been incomplete in various ways. Cooperation concerning RPL and
“active” nodes. This can be fulfilled by conserving energy to prolong TCP is demanded to provide a full-scale TCP for LLNs. Specifically, a
the lifetime of the network. According to the broadcast nature of striking research topic could be the interaction of multipath RPL
wireless channels, sensor nodes may hear transmissions, although approaches with TCP, because unordered packets usually affect TCP
they may not be the anticipated receivers of these transmissions. The throughput negatively due to fast-retransmit activities. According to
process of overhearing and receiving a transmission depletes a sig­ our statistics on TCP over RPL, there is one paper that handles this
nificant amount of energy in large networks (Redi and Basu, 2004). issue and particularly confirms that TCP expenses substantial
Moreover, gratuitously high-power transmissions consume energy throughput to preserve its reliability. Moreover, TCP suffers from
(Panichpapiboon et al., 2009), which reduces the network’s lifetime throughput unfairness among nodes (Kim et al., 2015b).
and generates interference in shared wireless mediums. Prior • Detection and control of congestion efforts: As aforementioned,
research considers adjusting transmission ranges in two ways; the RPL methods to detect and alleviate congestion are not mature. As a
first is determining the minimum transmission range required to result, RPL requires congestion detection procedures to effectively

47
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

support the routing in LLNs. Regulating workload rates is one solu­ challenge, several research papers have studied the characteristics of
tion to relieve congestion, but it negatively impacts latency. Another these networks and tried to suggest IoT routing protocols suitable for the
solution is that if any node along the route is detected to be con­ nature of LLNs. The ROLL WG has identified RPL as the standard routing
gested, the traffic must be re-routed. Specifically, other nodes, as protocol for IoT networks. However, many studies have indicated that
well, should detect congestion before re-routing traffic. Besides, the RPL has severe limitations and shortcomings. Hence, in recent years,
link loss rate requires consideration while modeling congestion various new solutions have emerged, which have been studied in the
detection methods. Consequently, in RPL, the routing decision context of this survey, in an attempt to mitigate existing routing prob­
should be determined by considering the congestion degree of the lems. This survey paper reviews the most widely used IoT routing pro­
preceding nodes, which is estimated based on the network’s tocol (i.e., RPL). It also presents a systematic review of RPL
connection and dynamics. specifications, recent RPL applications, and the latest RPL simulation
• Design of the cross-layer protocol: As previously stated, some platforms. Moreover, it details many recent RPL-based routing protocols
modifications to IEEE 802.15.4 are required to ensure improved for IoT/LLNs, which were designed to solve problems associated with
performance over RPL. Besides, low-duty cycle issues may impact the efficient routing support, mobility support, energy efficiency, cross-
performance where RPL constructs a DODAG and IEEE 802.15.4 layer issues, congestion detection and control, workload balancing,
enables the design of a cluster-tree graph. Many research papers have diversification of OFs, QoS, and security. Accordingly, our comprehen­
recently been published to investigate the cross-layer design of IEEE sive analysis of the studied routing protocols will certainly motivate
802.15.4 and RPL. In (Pavkovic et al., 2014), the authors suggested a researchers to develop cutting-edge solutions as particularly various
strategy for exploiting both RPL and IEEE 802.15.4 in topology limitations and negative aspects of the current proposals and solutions
construction. Another cross-layer-based design was proposed by are completely clarified. As a result, the authors’ insights, future direc­
Theoleyre et al., which is basically a modified MAC protocol that tion, and recommendations are highlighted to be taken into account
relies on IEEE 802.15.4. It intends to achieve high throughput by when designing new RPL-based solutions. After studying different
supporting RPL and authorizing nodes to get multiple parents numbers of simulators, we conclude that COOJA is the most applicable
simultaneously (Pavković et al., 2011). RPL is planned for an LLN, simulator for RPL because it has been fully implemented and is used in
which is frequently a standalone network. As a result, according to the preponderance of research papers. It is worth noting that the NS-3
(Kim et al., 2017a), cross-layer design and performance enhance­ simulator is almost ready. However, despite being fully developed, it
ment are not optional, but rather required to use RPL in reality. As seems that the RPL is still not operating properly.
previously mentioned, cross-layer-based approaches relying on RPL
were adopted by several studies to control jamming and achieve load Credit author statement
balancing. The concentration of RPL approaches relies on how the
congestion is controlled, which is based on each hop alone. Most of Khalid A. Darabkh and Muna Al-Akhras: Conceptualization,
these approaches interoperate with the lower layers. By contrast, methodology, and formal analysis. Khalid A. Darabkh, Muna Al-
end-to-end CC is enhanced in RPL by interacting with the transport Akhras, Jumana N. Zomot: Investigation as well as writing-revising
layer. For instance, multi-hop WSNs’ TCP is demanded to treat and editing the article. Khalid A. Darabkh, Muna Al-Akhras,
packet losses differently based on the source of losses such as Jumana N. Zomot, and Mohammed Atiquzzaman: reviewing and
congestion, channel noise, and mobility. Thus, the cross-layer design editing the article.
issue involving TCP and RPL needs to be investigated.
• Memory and power aware routing protocols: Unfortunately, most Declaration of competing interest
of the currently proposed RPL protocols tend to be computationally
heavy and complicated, thereby requiring higher storage and pro­ The authors declare that they have no known competing financial
cessing capabilities, which go against the constraints of IoT nodes. interests or personal relationships that could have appeared to influence
Besides this, the COOJA simulator motes have limited flash memory. the work reported in this paper.
For instance, Sky mote has a flash memory of 48 KB ROM, Z1 mote
has a flash memory of 96 KB ROM, and Wismote has a flash memory Data availability
of 256 KB ROM, which give rise for the researchers to pay extra
attention to the computational complexity influenced by their sug­ No data was used for the research described in the article.
gested ideas. In a nutshell, the studied RPL routing solutions have
outlined that the foremost constraints of the state-of-the-art routing Acknowledgments
solutions are the large complexity, induced by the employment of
optimization methods (like Fuzzy logic), a dramatical increase in The authors would like to acknowledge the Deanship of Scientific
DIO message length, excessive use of control messages, particularly, Research at the University of Jordan for the financial support granted for
at mobility usage, and, finally, the implementation overheads. this work.
• Investigating the security challenge: RPL security is a major
concern that must be investigated. In addition, new solutions must be References
proposed to address the RPL attacks and their effects on the network,
bearing in mind the previous solutions and their drawbacks. [Online]. Available: 3 July 2019. http://wsim.gforge.inria.fr/overview.html.
Abdel Fadeel, K.Q., El Sayed, K., 2015. ESMRF: enhanced stateless multicast RPL
forwarding for IPv6-based low-power and lossy networks. In: The 2015 Workshop on
9. Conclusion IoT Challenges in Mobile and Industrial Systems. Italy, Florence.
Abdel Hakeem, S., Hady, A., Kim, H., 2019. RPL routing protocol performance in smart
The advent of the IoT motivated the IETF to create the 6LoWPAN WG grid applications based wireless sensors: experimental and simulated analysis.
Electronics. 8 (2), 186.
to develop a solution for using IPv6 within LLNs, which consist of low- Abomhara, M., Køien, G.M., 2015. Cyber security and the internet of things. Journal of
power devices that are deployed in a wide range of IoT applications. Cyber Security. 4 (1), 65–88.
Since routing in networks with power-limited IoT devices is a major Abreu, C., Ricardo, M., Mendes, P.M., 2014. Energy-aware routing for biomedical
wireless sensor networks. J. Netw. Comput. Appl. 40, 270–278.

48
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Adat, V., Gupta, B.B., 2018. Security in Internet of Things: issues, challenges, taxonomy, Bouaziz, M., Rachedi, A., Belghith, A., Berbineau, M., Al-Ahmadi, S., August 2019. EMA-
and architecture. Telecommun. Syst. 67, 423–441. RPL: energy and mobility aware routing for the internet of mobile things. Future
Aijaz, A., Aghvami, H., 2015. Cognitive machine-to-machine communications for Generat. Comput. Syst. 97, 247–258.
internet-of-things: a protocol stack perspective. IEEE Internet Things J. 2 (2), Brandt, A., Buron, J., Porcu, G., April 2010. Home automation routing requirements in
103—112. low-power and lossy networks. IETF, RFC 5826, 1–16.
Aijaz, A., Su, H., Aghvami, A.-H., 2015. CORPL: a routing protocol for cognitive radio Cao, Y., Muqing, W., 2018. A novel RPL algorithm based on chaotic genetic algorithm.
enabled AMI networks. IEEE Trans. Smart Grid 6 (1), 477–485. Sensors 18 (11), 3647.
Airehrour, D., Gutierrez, J., Ray, S.K., 2016. Securing RPL routing protocol from Capone, S., Brama, R., Accettura, N., 2014a. An energy efficient and reliable composite
blackhole attacks using a trust-based mechanism. In: 26th International metric for RPL organized networks. In: IEEE 2014 12th IEEE International
Telecommunication Networks and Applications Conference (ITNAC). Dunedin, New Conference on Embedded and Ubiquitous Computing. EUC), Milano, Italy.
Zealand. Capone, S., Brama, R., Accettura, N., Striccoli, D., Boggia, G., 2014b. An energy efficient
Airehrour, D., Gutierrez, J., Ray, S., 2017. A trust-aware RPL routing protocol to detect and reliable composite metric for RPL organized networks. In: IEEE 2014 12th IEEE
blackhole and selective forwarding attacks. Australian Journal of International Conference on Embedded and Ubiquitous Computing. EUC), Milano,
Telecommunications and the Digital Economy. 5, 50–69. Italy.
Al-Fuqaha, A., Guizani, M., Mohammadi, M., Aledhari, M., Ayyash, M., 2015. Internet of Carneiro, G., 2010. Ns-3: network simulator 3. UTM Lab Meeting. 20.
things: a survey on enabling technologies, protocols and applications. IEEE Castellani, A.P., Rossi, M., Zorzi, M., 2014. Back pressure congestion control for CoAP/
Communications Surveys & Tutorials. 17, 2347–2378. 6LoWPAN networks. Ad Hoc Netw. 18, 71–84.
Al-Kashoash, H., Hassen, F., Kharrufa, H., Kemp, A.H., 2017a. Analytical Modelling of Charles, A.S.J., Kalavathi, P., 2022. A reliable link quality-based RPL routing for Internet
Congestion for 6LoWPAN Networks. ICT Express 4 (4), 209–215. of Things. Soft Comput. 26 (1), 123–135.
Al-Kashoash, H., Hafeez, M., Kemp, A., 2017b. Congestion control for 6LoWPAN Chen, Y., Chanet, J.-P., Hou, K.-M., 2012. RPL Routing Protocol a case study: precision
networks: a game theoretic framework. IEEE Internet Things J. 4 (3), 760–771. agriculture. In: First China-France Workshop on Future Computing Technology. CF-
Al-Kashoash, H., Amer, H., Mihaylova, L., Kemp, A.H., 2017c. Optimization-based hybrid WoFUCT 2012), Harbin, China.
congestion alleviation for 6LoWPAN networks. IEEE Internet Things J. 4 (6), Chen, Yibo, Chanet, Jean-Pierre, Hou, Kun-Mean, Shi, Hongling, Sousa, Gil De, et al.,
2070–2081. 2015. A scalable context-aware objective function (SCAOF) of routing protocol for
Alaa, M., Zaidan, A.A., Zaidan, B.B., Talal, M., Kiah, M.L., 2017. A review of smart home agricultural low-power and lossy networks (RPAL). Sensors 15 (8), 19507–19540.
applications based on Internet of Things. J. Netw. Comput. Appl. 97 (1), 48–65. Chen, Y., Hou, K.-M., Chanet, J.-P., el Gholami, K., 2013. A RPL based Adaptive and
Alhasanat, M., Althunibat, S., Darabkh, K.A., Alhasanat, A., Alsafasfeh, M., 2020. Scalable Data-collection Protocol module for NS-3 simulation platform. In:
A physical-layer key distribution mechanism for IoT networks. Mobile Network. NICST’2103 New Information Communication Science and Technology for
Appl. 25, 173–178. Sustainable Development: France-China International Workshop. Clermont-Ferrand,
Almusaylim, Z.A., Alhumam, A., Jhanjhi, N.Z., 2020. Proposing a secure rpl based France.
internet of things routing protocol: a review. Ad Hoc Netw. 101 (6), 102096. Chen, N., Qiu, T., Daneshmand, M., Wu, D.O., 2021. Robust networking: dynamic
Alsukayti, I.S., Alreshoodi, M., 2019. Toward an understanding of recent developments topology evolution learning for internet of things. ACM Trans. Sens. Netw. 17 (3),
in RPL routing. IET Netw. 8 (6), 356–366. 1–23.
Anand, M.C.R., Tahiliani, M.P., 2016. mRPL++: smarter-HOP for optimizing mobility in Chen, N., Qiu, T., Lu, Z., Wu, D.O., 2022. An adaptive robustness evolution algorithm
RPL. In: IEEE 2016 IEEE Region 10 Symposium (TENSYMP ). Bali, Indonesia. with self-competition and its 3D deployment for internet of things. IEEE/ACM Trans.
Ancillotti, E., Bruno, R., Conti, M., 2014a. Reliable data delivery with the IETF routing Netw. 30 (1), 368–381.
protocol for low-power and lossy networks. IEEE Trans. Ind. Inf. 10 (3), 1864–1877. Cobarzan, C., Montavont, J., Noel, T., 2014a. Analysis and performance evaluation of
Ancillotti, E., Bruno, R., Conti, M., Mingozzi, E., Vallati, C., 2014b. Trickle-L2: RPL under mobility. In: IEEE 2014 IEEE Symposium on Computers and
lightweight link quality estimation through Trickle in RPL networks. In: IEEE 15th Communication (ISCC). Funchal, Madeira, Portugal.
International Symposium on "A World of Wireless, Mobile and Multimedia Cobarzan, C., Montavont, J., Noel, T., 2014b. Analysis and performance evaluation of
Networks. WoWMoM), Sydney, Australia. RPL under mobility. In: IEEE 2014 IEEE Symposium on Computers and
Araújo, H., Filho, R., Rodrigues, J., Rabelo, R., Sousa, N., Filho, J., Sobral, J., 2018. Communication (ISCC). Funchal, Madeira, Portugal.
A proposal for IoT dynamic routes selection based on contextual information. Council, N., 2008. Disruptive civil technologies: six technologies with potential impacts
Sensors. 18 (2), 353. on us interests out to 2025. In: Conference Report CR. SRI Consulting Business
Aris, A., Oktug, S.F., Yalcin, S.B.O., 2016. RPL version number attacks: in-depth study. Intelligence, Menlo Park, CA, USA.
In: NOMS 2016 - IEEE/IFIP Network Operations and Management Symposium. Cuomo, F., Abbagnale, A., Cipollone, E., 2013. Cross-layer network formation for energy-
Istanbul, Turkey. efficient IEEE 802.15.4/ZigBee wireless sensor networks. Ad Hoc Netw. 11 (2),
Arslan, M., Bin Zikria, Y., Zulqarnain, Sung Won, K., 2020. Routing protocol for Low- 672–686.
Power and Lossy Networks for heterogeneous traffic network. EURASIP J. Wirel. Darabkh, K.A., Al-Akhras, M., 2021. RPL over internet of things: challenges, solutions,
Commun. Netw. 2020 (1), 21. and recommendations. In: 2021 IEEE International Conference on Mobile Networks
Avila, K., Jabba, D., Gomez, J., September 2020. Security aspects for rpl-based protocols: and Wireless Communications (ICMNWC). Sri Siddhartha Institute of Technology,
a systematic review in IoT. Appl. Sci. 10 (8), 6472. Tumakuru, Karnataka, India.
Baccelli, E., Philipp, M., Goyal, M., 2011. The P2P-RPL routing protocol for IPv6 sensor Darabkh, K.A., Al-Jdayeh, L., 2019. AEA-FCP: an adaptive energy-aware fixed clustering
networks: testbed experiments. In: SoftCOM 2011, 19th International Conference on protocol for data dissemination in wireless sensor networks. Personal Ubiquitous
Software. Telecommunications and Computer Networks, Split, Croatia. Comput. 23 (5), 819–837.
Badach, A., 2018. RPL routing metrics and constraints. In: Project: Internet of Things – Darabkh, K.A., Amro, O.M., 2020. New routing protocol for half-duplex cognitive radio
Technologies. Protocols and Applications, pp. 1–4. ad-hoc networks over IoT environment. In: 2020 IEEE International IOT, Electronics
Bang, A.O., Rao, U.P., Kaliyar, P., Conti, M., 2022. Assessment of routing attacks and and Mechatronics Conference (IEMTRONICS). Canada, Vancouver, BC.
mitigation techniques with rpl control messages: a survey. ACM Comput. Surv. 55 Darabkh, K.A., Kassab, W.K., 2020. Maximizing the life time of wireless sensor networks
(2), 1–36. over IoT environment. In: 2020 Fifth International Conference on Fog and Mobile
Banh, M., Nguyen, N., Phung, K.-H., Nguyen, L., Thanh, N.H., Steenhaut, K., 2016. Edge Computing. France, Paris.
Energy balancing RPL-based routing for internet of things. In: IEEE Sixth International Darabkh, K.A., Albtoush, W.Y., Jafar, I.F., May 2017. Improved clustering algorithms for
Conference On Communications And Electronics (ICCE), Ha-Long City. Quang Ninh target tracking in wireless sensor networks. J. Supercomput. 73 (5), 1952–1977.
Province, Vietnam. Darabkh, K.A., El-Yabroudi, M.Z., El-Mousa, A.H., 2019a. BPA-CRP: a balanced power-
Barcelo, M., Correa, A., Vicario, J.L., Morell, A., Vilajosana, X., 2016. Addressing aware clustering and routing protocol for wireless sensor networks. Ad Hoc Netw.
mobility in RPL with position assisted metrics. IEEE Sensor. J. 16 (7), 2151–2161. 82, 155–171.
Barthel, D., Vasseur, J., Pister, K., Kim, M., Dejean, N., 2012. Routing metrics used for Darabkh, K.A., Odetallah, S.M., Al-qudah, Z., Khalifeh, A.F., Shurman, M.M., 2019b.
path calculation in low-power and lossy networks. IETF, RFC 6551, 1–30. Energy–aware and density-based clustering and relaying protocol (EA-DB-CRP) for
Bartolozzi, L., Pecorella, T., Fantacci, R., 2012. ns-3 RPL module: IPv6 routing protocol gathering data in wireless sensor networks. Appl. Soft Comput. 80, 154–166.
for low power and lossy networks. In: Proceedings of the 5th International ICST Darabkh, K.A., Zomot, J.N., Al-qudah, Z., Edb-Chs, B.O.F., 2019c. Energy and distance
Conference on Simulation Tools and Techniques. Desenzano del Garda, Italy. based cluster head selection with balanced objective function protocol. IET
Belghachi, M., Feham, M., 2015. QoS routing RPL for low power and lossy networks. Int. Communications, Special Issue: Future of Intelligent Wireless LANs 13 (19),
J. Distributed Sens. Netw. 1–10. 3168–3180.
Ben Aissa, Y., Grichi, H., Khalgui, M., Koubâa, A., Bachir, A., 2019. QCOF: new RPL Darabkh, K.A., Kassab, W.K., Khalifeh, A.F., LiM-Ahp, -G.-C., 2020a. Life time
extension for QoS and congestion-aware in low power and lossy network. In: 14th maximizing based on analytical hierarchal process and genetic clustering protocol
International Conference on Software Technologies (ICSOFT 2019). Prague, Czech for the internet of things environment. Comput. Network. 176, 107253.
Republic. Darabkh, K.A., Amro, O.M., Al-Zubi, R.T., Bany Salameh, H., Saifan, R., 2020b. JavaSim-
Ben Saad, L., Chauvenet, C., Tourancheau, B., 2011. Simulation of the RPL routing IBFD-CRNs: novel Java simulator for in-band full-duplex cognitive radio networks
protocol for IPv6 sensor networks: two cases studies. In: International Conference on over internet of things environment. J. Netw. Comput. Appl. 172, 102833.
Sensor Technologies and Applications SENSORCOMM 2011. France, Nice. Darabkh, K.A., Amro, O.M., Al-Zubi, R.T., Bany Salameh, H., 2021a. Yet efficient routing
Borlase, S., 2013. Smart Grids Infrastructure, Technology, and Solutions. CRC Press. protocols for half- and full-duplex cognitive radio ad-hoc networks over IoT
Bormann, C., Castellani, A.P., Shelby, Z., 2012. CoAP: an application protocol for billions environment. J. Netw. Comput. Appl. 173, 102836.
of tiny internet nodes. IEEE Internet Computing 16 (2), 62–67. Darabkh, K.A., Al-Akhras, M., Khalifeh, A., Oct 2021. Improving routing protocol for
Bouaziz, M., Rachedi, A., Belghith, A., EKF-MRPL, April 2019. Advanced mobility low-power and lossy networks over IoT enviroment. In: Proceedings of the 30th IEEE
support routing protocol for internet of mobile things: movement prediction Wireless and Optical Communications Conference (WOCC 2021). Taipei Tech,
approach. Future Generat. Comput. Syst. 93, 822–832. Taipei, Taiwan.

49
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Darabkh, K.A., Zomot, J.N., Al-qudah, Z., Khalifeh, A.F., 2022a. Impairments-aware time Idrees, A.K., Witwit, A.J.H., 2021. Energy-efficient load-balanced RPL routing protocol
slot allocation model for energy-constrained multi-hop clustered IoT nodes for internet of things (IoTs) networks. Int. J. Internet Technol. Secur. Trans. 11 (3),
considering TDMA and DSSS MAC protocols. Journal of Industrial Information 286–306.
Integration 25, 100243. Iova, O., Theoleyre, F., Noel, T., 2013. Stability and efficiency of RPL under realistic
Darabkh, K.A., Al-Akhras, M., Khalifeh, A.F., Jafar, I.F., Jubair, F., 2022b. An innovative conditions in wireless sensor networks. In: IEEE 2013 IEEE 24th Annual
RPL objective function for broad range of IoT domains utilizing fuzzy logic and International Symposium on Personal, Indoor, and Mobile Radio Communications.
multiple metrics. Expert Syst. Appl. 205, 117593. PIMRC), London, United Kingdom.
Dhumane, A., Prasad, R., 2015. Routing challenges in internet of things. CSI Iova, O., Theoleyre, F., Noel, T., 2015. Using multiparent routing in RPL to increase the
Communications Magazine 3, 19–20. stability and the lifetime of the network. Ad Hoc Netw. 29, 45–62.
Dohler, M., Watteyne, T., Winter, T., Barthel, D., May 2009. Routing requirements for Iova, O., Picco, P., Istomin, T., Kiraly, C., 2016. RPL: the routing standard for the internet
urban low-power and lossy networks. IETF, RFC 5548, 1–20. of things... Or is it? IEEE Commun. Mag. 54 (12), 16–22.
Evans, D., 2011. The internet of things—how the next evolution of the internet is Issariyakul, T., Hossain, E., 2012. Introduction to Network Simulator NS2. Springer, New
changing everything. CISCO white paper 1, 1–11. York.
Feroz, A., Ikram, A.-u.-A., 2020. RPL objective functions evaluation for the application of Jadhao, A., Solapure, S., 2017. Analysis of routing protocol for low power and lossy
smart grid AMI networks. Am. J. Comput. Sci. Inf. Technol. 8 (2). networks (RPL) using cooja simulator. In: International Conference on Wireless
Fotouhi, H., Moreira, D., Alves, M., 2015. mRPL: boosting mobility in the internet of Communications, Signal Processing and Networking. WiSPNET), Chennai, India.
things. Ad Hoc Netw. 26, 17–35. Jadhav, R., Sahoo, R., Cao, Z., 2017. No-Path DAO Modifications. ROLL [Online].
Gaddour, O., Koubâa, A., 2012. RPL in a nutshell: a survey. Comput. Network. 56 (14), Available:April 2021. https://tools.ietf.org/id/draft-ietf-roll-efficient-npdao-00.
3163–3178. html.
Gaddour, O., Koubaa, A., Baccour, N., Abid, M., 2014. OF-FL: QoS-aware fuzzy logic Janicijevic, N., Lukic, M., Mezei, I., 2011. Routing protocol for low-power and Lossy
objective function for the RPL routing protocol. In: IEEE 2014 12th International wireless sensor networks. In: IEEE 2011 19th Telecommunications Forum Telfor
Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless (TELFOR)Proceedings of Papers. Belgrade,Serbia.
Networks (WiOpt). Hammamet, Tunisia. Jarry, A., Leone, P., Powell, O., Rolim, J., 2006. An optimal data propagation algorithm
Gara, F., Ben Saad, L., Ben Ayed, R., Tourancheau, B., 2015. RPL protocol adapted for for maximizing the lifespan of sensor networks. In: Distributed Computing in Sensor
healthcare and medical applications. In: IEEE 2015 International Wireless Systems. Springer, pp. 405–421.
Communications and Mobile Computing Conference. IWCMC), Dubrovnik, Croatia. Javaid, N., Faisal, S., Khan, Z.A., Nayab, D., Mehmood, Z., 2013. Measuring fatigue of
Gardasevic, G., Mijovic, S., Stajkic, A., Buratti, C., 2015. On the performance of soldiers in wireless body area sensor networks. In: IEEE 2013 Eighth International
6LoWPAN through experimentation. In: IEEE 2015 International Wireless Conference on Broadband and Wireless Computing, Communication and
Communications and Mobile Computing Conference (IWCMC). Dubrovnik, Croatia. Applications (BWCCA). Compiegne, France.
Gastón Lorente, G., Lemmens, B., Carlier, M., Braeken, A., Steenhaut, K., 2017. BMRF: Jun, L., Mianxiong, D., Ota, K., Anfeng, L., Songyuan, H., 2015. Reliability guaranteed
bidirectional multicast RPL forwarding. Ad Hoc Netw. 54, 69–84. efficient data gathering in wireless sensor networks. IEEE Access 3, 430–444.
Gaur, P., Tahiliani, M.P., 2015. Operating systems for IoT devices: a critical survey. In: Kamble, A., Malemath, V.S., Patil, D., 2017. Security attacks and secure routing protocols
IEEE Region 10 Symposium (TENSYMP). Ahmedabad, India. in RPL-based internet. In: International Conference on Emerging Trends &
Ghaleb, B., Al-Dubai, A., Ekonomou, E., Alsarhan, A., Nasser, Y., Mackenzie, L., Innovation in ICT (ICEI). Pune, India.
Boukerche, A., 2019. A survey of limitations and enhancements of the IPv6 routing Kamgueu, P.O., Nataf, E., Djotio, T., Festor, O., 2012. Energy-based Routing Metric for
protocol for low-power and lossy networks: a focus on core operations. IEEE RPL. Ph.D. dissertation, INRIA.
Communications Surveys & Tutorials 21 (2), 1607–1635. Kamgueu, P.-O., Nataf, .E., Ndie Djotio, T., 2015. On design and deployment of fuzzy-
Gnawali, O., Levis, P., 2012. The minimum rank with hysteresis objective function. IETF, based metric for routing in low-power and lossy networks. In: IEEE 2015 IEEE 40th
RFC 6719, 1–13. Local Computer Networks Conference Workshops. LCN Workshops), Clearwater
Goyal, M., Baccelli, E., Philipp, M., Brandt, A., Martocci, A., 2013. Reactive Discovery of Beach, FL, USA.
Point-to-Point Routes in Low-Power and Lossy Networks. IETF, RFC 6997, 1–40. Kassab, W., Darabkh, K.A., 2020. A-Z survey of internet of things: architectures,
Gozuacik, N., Oktug, S.F., 2015. Parent-aware routing for IoT networks. In: Internet of protocols, applications, recent advances, future directions and recommendations.
Things, Smart Spaces, and Next Generation Networks and Systems, 15th J. Netw. Comput. Appl. 163, 102663.
International Conference, NEW2AN 2015, and 8th Conference. ruSMART 2015, St. Kaviani, F., Soltanaghaei, M., 2022. CQARPL: congestion and QoS-aware RPL for IoT
Petersburg, Russia. applications under heavy traffic. J. Supercomput.
Granjal, J., Monteiro, E., Silva, J.S., 2015. Security for the internet of things: a survey of Khalifeh, A., Aldahdouh, K.A., Darabkh, K.A., Al-Sit, W., 2019. A survey of 5G emerging
existing protocols and open research issues. IEEE Communications Surveys & wireless technologies featuring LoRaWAN, sigfox, NB-IoT and LTE-M. In:
Tutorials 17 (3), 1294–1312. International Conference on Wireless Communications Signal Processing and
Gupta, N., Pughat, A., Sharma, V., 2021. A critical analysis of RPL objective functions in Networking. WiSPNET), Chennai, India.
internet of things paradigm. Peer-to-Peer Networking and Applications 14 (11), Khalifeh, A., Darabkh, K.A., Khasawneh, A., Alqaisieh, I., Alrubaye, S., AlAbdala, A.,
2187–2208. Salameh, M., Alassaf, A., Al-HajAli, S., Al-Wardat, R., Bartolini, N., Bongiovanni, G.,
Gutierrez, J.A., Naeve, M., Callaway, E., Bourgeois, M., Mitter, V., Heile, B., 2001. IEEE Rajendiran, K., 2021. Wireless sensor networks for smart cities: network design,
802.15.4: a developing standard for low-power low-cost wireless personal area implementation and performance evaluation. Electronics 10 (2), 218.
networks. IEEE Network 15 (5), 12–19. Khan, A.u.R., Bilal, S.M., Othman, M., 2012. A performance comparison of open source
Ha, M., Kwon, K., Kim, D., Kong, P.-Y., 2014. Dynamic and distributed load balancing network simulators for wireless networks. In: IEEE 2012 IEEE International
scheme in multi-gateway based 6LoWPAN. In: IEEE International Conference on Conference on Control System, Computing and Engineering (ICCSCE). Penang,
Internet of Things(iThings), and IEEE Green Computing and Communications Malaysia.
(GreenCom) and IEEE Cyber, Physical and Social Computing. CPSCom), Taipei, Khan, F.I., Shon, T., Lee, T., Kim, K., 2013. Wormhole attack prevention mechanism for
Taiwan. RPL based LLN network. In: Fifth International Conference on Ubiquitous and Future
Harb, H., Idrees, A.K., Jaber, A., Makhoul, A., Zahwe, O., Taam, M., 2017. Wireless Networks (ICUFN). Da Nang.
sensor networks: a big data source in internet of things. Int. J. Sensor. Wireless Khan, M.M., Lodhi, A., Rehman, A., Khan, A., Hussain, F.B., 2016. Sink-to-Sink
Commun. Control 7 (2), 93–109. coordination framework using RPL: routing protocol for low power and lossy
Harbi, Y., Aliouat, Z., Refoufi, A., Harous, S., 2021. Recent security trends in internet of networks. J. Sens. 1–11.
things: a comprehensive survey. IEEE Access. 9, 113292–113314. Kharrufa, H., Al-Kashoash, H., Al-Nidawi, Y., Mosquera, M.Q., Kemp, A.H., 2017.
Hargreaves, T., Charlie, W., Hauxwell-Baldwin, R., 2017. Learning to live in a smart Dynamic RPL for multi-hop routing in IoT applications. In: IEEE 2017 13th Annual
home. Build. Res. Inf. 46 (1), 127–139. Conference on Wireless On-Demand Network Systems and Services (WONS),
Hellaoui, H., Koudil, M., 2015. Bird flocking congestion control for CoAP/RPL/ Jackson. WY, USA.
6LoWPAN networks. In: The 2015 Workshop on IoT Challenges in Mobile and Kharrufa, H., Al-Kashoash, H., Kemp, A.H., April 2019. RPL-based routing protocols in
Industrial Systems - IoT-Sys, vol. 15. ACM Press the 2015 Workshop), Florence, Italy. IoT applications: a review. IEEE Sensor. J. 19 (15), 5952–5967.
Homoliak, I., Toffalini, F., Guarnizo, J., Elovici, Y., Ochoa, M., 2019. Insight into insiders Khelifi, N., Oteafy, S., Hassanein, H., Youssef, H., 2015. Proactive maintenance in RPL for
and it: a survey of insider threat taxonomies, analysis, modeling, and 6LowPAN. In: IEEE 2015 International Wireless Communications and Mobile
countermeasures. ACM Comput. Surv. 52 (2), 1–40. Computing Conference (IWCMC). Dubrovnik, Croatia.
Hong, K.-S., Choi, L., 2011. DAG-based multipath routing for mobile sensor networks. In: Kim, Y., Paek, J., 2020. NG-RPL for efficient P2P routing in low-power multihop wireless
IEEE 2011 International Conference on ICT Convergence (ICTC). Seoul, Korea. networks. IEEE Access. 8, 182591–182599.
Hui, J., Kelsey, R., 2016. Multicast Protocol for Low-Power and Lossy Networks (MPL), Kim, H.-S., Paek, J., Bahk, S., Qu-Rpl, 2015a. Queue utilization based RPL for load
IETF. RFC 7731, 1–28. balancing in large scale industrial applications. In: IEEE 2015 12th Annual IEEE
Hwang, R.-H., Peng, M.-C., Wu, C.-Y., Abimannan, S., 2020. A novel RPL-based multicast International Conference on Sensing, Communication, and Networking. SECON) -
routing mechanism for wireless sensor networks. Int. J. Ad Hoc Ubiquitous Comput. Seattle, WA, USA.
33 (2), 122–131.

50
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Kim, H.-S., Im, H., Lee, M.-S., Paek, J., Bahk, S., 2015b. A measurement study of TCP Lu, Y., Morris, K., Frechette, S., 2019. Current Standards Landscape for Smart
over RPL in low-power and lossy networks. J. Commun. Network. 17 (6), 647–655. Manufacturing Systems. , [Online]. Available: January 2016. https://nvlpubs.nist.
Kim, H.-S., Kim, H., Paek, J., Bahk, S., 2016. Load balancing under heavy traffic in RPL gov/nistpubs/ir/2016/NIST.IR.8107.pdf.
routing protocol for low power and lossy networks. IEEE Trans. Mobile Comput. 16 Ma, C., Sheu, J.-P., Hsu, C.-X., 2016. A game theory based congestion control protocol for
(4), 964–979. wireless personal area networks. J. Sens. 1–13.
Kim, H.S., Ko, J., Culler, D.E., Paek, J., 2017a. Challenging the IPv6 routing protocol for Maisonneuve, N., Stevens, M.L., Niessen, M., Hanappe, P., Steels, L., 2009. Citizen noise
low-power and lossy networks (rpl): a survey. IEEE Communications Surveys & pollution monitoring. In: The 10th Annual International Conference on Digital
Tutorials 19 (4), 2502–2525. Government Research, Partnerships for Public Innovation. Puebla, Mexico.
Kim, H.-s., Cho, H., Kim, H., Bahk, S., Dt-Rpl, 2017b. Diverse bidirectional traffic Martocci, J., Mil, P.D., Riou, N., Vermeylen, W., 2010. Building automation routing
delivery through RPL routing protocol in low power and lossy networks. Comput. requirements in low-power and lossy networks. IETF, RFC 5867, 1–25.
Network. 126 (24), 150–161. Mayzaud, A., Badonnel, R., Chrisment, I., 2016. A taxonomy of attacks in RPL-based
Kim, H.-s., Paek, J., Culler, D.E., Bahk, S., Rpl, P.C.-, 2020. Joint control of routing internet of things. Int. J. Netw. Secur. 18 (3), 459–473.
topology and transmission power in real low-power and lossy networks. ACM Trans. Mehmood, T., 2019. COOJA Network Simulator: Exploring the Infinite Possible Ways to
Sens. Netw. 16 (2), 1–32. Compute the Performance Metrics of IOT Based Smart Devices to Understand the
Kniess, J., Marques, V.d.F., 2020. MARPL: a crosslayer approach for Internet of things Working of IOT Based Compression & Routing Protocols. July 22 2017 [Online].
based on neighbor variability for mobility support in RPL. Transactions on Emerging Available: https://arxiv.org/abs/1712.08303.
Telecommunications Technologies 31 (12), e3931. Meredith, S., 2022. Two-thirds of Global Population Will Live in Cities by 2050, UN Says,
Ko, J., Jeong, J., Park, J., Jun, J.A., Omprakash, G., Paek, J., 2015. DualMOP-RPL: 15 March [Online]. Available:17 May 2018. https://www.cnbc.com/2018
supporting multiple modes of downward routing in a single RPL network. ACM /05/17/two-thirds-of-global-population-will-live-in-cities-by-2050-un-says.html.
Trans. Sens. Netw. 11 (2), 1–20. Michopoulos, V., Guan, L., Oikonomou, G., Phillips, I., 2012. DCCC6: duty Cycle-aware
Korbi, I.E., Ben Brahim, M., Adjih, C., 2012. Mobility enhanced RPL for wireless sensor congestion control for 6LoWPAN networks. In: IEEE 2012 IEEE International
networks. In: IEEE 2012 Third International Conference on the Network of the Conference on Pervasive Computing and Communications Workshops (PerCom
Future (NOF). Tunis, Tunisia. Workshops). Lugano, Switzerland.
Kumar, A., Hariharan, N., Dcrl-Rpl, 2020. Dual context-based routing and load balancing Mishra, A.K., Singh, O., Kumar, A., Puthal, D., 2022. Hybrid Mode of Operations for RPL
in RPL for IoT networks. IET Commun. 14 (12), 1869–1882. in IoT: A Systematic Survey. IEEE Transactions on Network and Service
Lalani, S.R., Salehi, A., Taghizadeh, H., Safaei, B., Monazzah, A., Ejlali, A., 2020. REFER: Management, pp. 1–15.
a reliable and energy-efficient rpl for mobile IoT applications. In: IEEE 2020 CSI/ Misra, P., Ahmed, N., Ostry, D., Jha, S., 2011. Characterization of asymmetry in low-
CPSSI International Symposium on Real-Time and Embedded Systems and power wireless links: an empirical study. In: Distributed Computing and Networking.
Technologies. RTEST), Tehran, Iran. Springer, Berlin, Heidelberg, pp. 340–351.
Lamaazi, H., Benamar, N., 2019. A novel approach for RPL assessment based on the Mohamed, K., Ali, S., Ali, S., Kassim, I., 2021. Performance evaluation of RPL routes and
objective function and trickle optimizations. Wireless Commun. Mobile Comput. DODAGConstruction for IoTs. International Journal of Digital Society (IJDS) 12 (1),
1–9. 1–11.
Lamaazi, H., Benamar, N., 2020a. A comprehensive survey on enhancements and Mohammadinejad, H., Mohammadhoseini, F., 2019. Proposing a method for enhancing
limitations of the RPL protocol: a focus on the objective function. Ad Hoc Netw. 96 the reliability of RPL routing protocol in the smart grid neighborhood area networks.
(3), 102001. Int. J. Comput. Netw. Inf. Secur. 11 (7), 21–28.
Lamaazi, H., Benamar, N., 2020b. RPL enhancement based FL-trickle: a novel flexible Moradi, S., Javidan, R., 2020. A new objective function for RPL routing protocol in IoT to
trickle algorithm for low power and lossy networks. Wireless Pers. Commun. 110 (3), increase network lifetime. Int. J. Wireless Mobile Comput. 19 (1), 73–79.
1403–1428. Moritz, G., Lerche, T., 2013. Beyond 6LoWPAN: Web services in wireless sensor
Lamaazi, H., El Ahmadi, A., Benamar, N., Jara, A.J., Of-Ecf, 2019. A new optimization of networks. IEEE Trans. Ind. Inf. 9 (4), 1795–1805.
the objective. In: IEEE 2019 International Conference on Wireless and Mobile Mugdhe, S.P., 2022. Network Stack. ANGR, 15 June 2016 [Online]. Available: htt
Computing, Networking and Communications. WiMob), Barcelona, Spain. ps://anrg.usc.edu/contiki/index.php/Network_Stack.
Le, A., Loo, J., Luo, Y., Lasebae, A., 2013. The impacts of internal threats towards Narendra, K.S., Thathachar, M.A.L., 1974. Learning automata - a survey. IEEE
Routing Protocol for Low power and lossy network performance. In: IEEE Transactions on Systems Man and Cybernetics 4 (4), 323–334.
Symposium on Computers and Communications (ISCC). .. Split, Croatia. Narten, T., Nordmark, E., Simpson, W., Soliman, H., 2007. Neighbor Discovery for IP
Le, A., Loo, J., Chai, K.K., Aiash, M., 2016. A specification-based IDS for detecting attacks Version 6 (IPv6), IETF, RFC 4861, pp. 1–97.
on RPL-based network topology. Information 7 (2), 25. Nassar, J., Berthomé, M., Dubrulle, J., Gouvy, N., Mitton, N., 2018. Multiple instances
Lee, T., Lee, M.-S., Kim, H.-S., Bahk, S., 2016. A synergistic architecture for RPL over QoS routing in RPL: application to smart grids. Sensors 18 (8), 2472.
BLE. In: 13th Annual IEEE International Conference on Sensing, Communication, Niu, X., Huang, X., Zhao, Z., Zhang, Y., Huang, C., 2007. The design and evaluation of a
and Networking. SECON), London, UK. wireless sensor network for mine safety monitoring. In: IEEE GLOBECOM 2007 -
Lee, T.-H., Chang, L.-H., Liaw, J.-J., Shih, Y.-C., Hu, C.-L., Chu, H.-C., 2020. Service- IEEE Global Telecommunications Conference. Washington, DC, USA.
aware objective function with QoS for RPL routing. Int. J. Ad Hoc Ubiquitous Oh, S., Hwang, D., Kim, K., Kim, K., 2018. A hybrid mode to enhance the downward
Comput. 35 (1), 49–60. route performance in routing protocol for low power and lossy networks. Int. J.
Levis, P., Lee, N., 2003. TOSSIM: a simulator for TinyOS networks. UC Berkeley 24. Distributed Sens. Netw. 14 (4).
Levis, P., Patel, N., Culler, D., Shenker, S., 2004. Trickle: a self-regulating algorithm for Oikonomou, G., Phillips, I., 2012. Stateless multicast forwarding with RPL in 6LowPAN
code propagation and maintenance in wireless sensor networks. In: The First sensor networks. In: IEEE International Conference on Pervasive Computing and
Symposium on Networked Systems Design and Implementation. San Francisco, CA, Communications Workshops. Lugano, Switzerland.
USA. Oikonomou, G., Phillips, I., Tryfonas, T., 2013. IPv6 multicast forwarding in RPL-based
Levis, P., Madden, S., Polastre, J., Szewczyk, R., Whitehouse, K., Woo, A., Gay, D., wireless sensor networks. Wireless Pers. Commun. 73 (3), 1089–1116.
Hill, J., Welsh, M., Brewer, E., Culler, D., 2005. Ambient Intelligence || TinyOS: an Oliveira, A., Vazao, T., 2016. Low-power and lossy networks under mobility: a survey.
Operating System for Sensor Networks, 10.1007/b138670. Springer, Berlin, Comput. Network. 107 (2), 339–352.
pp. 115–148. [Online]. Available: https://www.nsnam.org/wiki/Installation.
Levis, P., Tavakoli, A., Dawson-Haggerty, S., 2009. Overview of Existing Routing [Online]. Available: https://www.nsnam.org/wiki/RoutesMobilityModel.
Protocols for Low Power and Lossy Networks. Internet Engineering Task Force. Palattella, M.R., Accettura, N., Vilajosana, X., Watteyne, T., Grieco, L.A., Boggia, G.,
Levis, P., Clausen, T., Hui, J., Gnawali, O., Ko, J., 2011. The Trickle Algorithm. IETF, RFC Dohler, M., 2013. Standardized protocol stack for the internet of (important) things.
6206, pp. 1–12. IEEE Communications Surveys & Tutorials 15 (3), 1389–1406.
Lim, C., 2019. A survey on congestion control for RPL-based wireless sensor networks. Pancaroglu, D., Sen, S., 2021. Load balancing for RPL-based internet of things: a review.
Sensors 19 (11), 2567. Ad Hoc Netw. 116, 102491.
Liu, A., Jin, X., Cui, G., Chen, Z., 2013a. Deployment guidelines for achieving maximum Panichpapiboon, S., Ferrari, G., Tonguz, O.K., 2009. Optimal transmit power in wireless
lifetime and avoiding energy holes in sensor network. Inf. Sci. 230, 197—226. sensor networks. IEEE Trans. Mobile Comput. 5 (10), 1432–1447.
Liu, X., Guo, J., Bhatti, G., Orlik, P., Parsons, K., 2013b. Load balanced routing for low Parkavi, T., Arockiam, D.L., 2021. A survey on Sinkhole attack in RPL. Annals of the
power and lossy networks. In: IEEE 2013 IEEE Wireless Communications and Romanian Society for Cell Biology 25 (1), 511–515.
Networking Conference. WCNC), Shanghai, China. Patil, A.K., Hadalgi, D., 2012. Evaluation of discrete event wireless sensor network
Liu, X., Sheng, Z., Yin, C., Ali, F., Roggen, D., 2017. Performance analysis of routing simulators. Intl.l Journal of Computer Science and Networks. 1 (5).
protocol for low power and lossy networks (RPL) in large scale networks. IEEE Pavković, B., Theoleyre, F., Duda, A., 2011. Multipath opportunistic RPL routing over
Internet Things J. 4 (6), 2172–2185. IEEE 802.15.4. ACM Press the 14th ACM international conference on Modeling,
Loh, K., Lynch, J.P., 2006. A summary review of wireless sensors and sensor networks for analysis and simulation of wireless and mobile systems - MSWiM 11. Miami, Florida,
structural health monitoring. Shock Vib. Digest 38 (2), 91–128. USA.
Pavkovic, B., Duda, A., Hwang, W., Theoleyre, F., 2014. Efficient topology construction
for rpl over IEEE 802.15. 4 in wireless sensor. Ad Hoc Netw. 15, 25–38.

51
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Perazzo, P., Vallati, C., Anastasi, G., Dini, G., 2017. DIO suppression attack against Sun, Y., Gurewitz, O., Du, S., Tang, L., Johnson, D.B., 2009. ADB: an efficient multihop
routing in the internet of things. IEEE Commun. Lett. 21 (11), 2524–2527. broadcast protocol based on asynchronous duty-cycling in wireless sensor networks.
Perkins, C., Johnson, D., Arkko, J., 2011. Mobility Support in IPv6, IETF, RFC 6275, In: The 7th ACM Conference on Embedded Networked Sensor Systems. California,
pp. 1–169. Berkeley.
Pister, K., Thubert, P., Dwars, S., Phinney, T., 2009. Industrial Routing Requirements in Taghizadeh, S., Bobarshad, H., Elbiaze, H., 2018. CLRPL: context-aware and load
Low-Power and Lossy Networks. IETF, RFC 5673, 1–26. balancing RPL for iot networks under heavy and highly dynamic load. IEEE Access 6,
Pongle, P., Chavan, G., 2015. A survey: attacks on RPL and 6LoWPAN in IoT. In: 23277–23291.
International Conference on Pervasive Computing (ICPC). Pune, India. Tahir, Y., Yang, S., McCann, J., 2018. BRPL: backpressure RPL for high-throughput and
Prasad, H., Babu, S., 2017. A survey on network routing protocols in internet of things mobile IoTs. IEEE Trans. Mobile Comput. 17 (1), 29–43.
(IOT). Int. J. Comput. Appl. 160 (2), 18–22. Talwar, M., 2015. Routing techniques and protocols for internet of things: a survey. In:
Pu, C., 2018. Mitigating DAO inconsistency attack in RPL-based low power and lossy Proceeding of NCRIET-2015 & Indian. J.Sci.Res., Bidar.
networks. In: IEEE 8th Annual Computing and Communication Workshop and Tang, W., Ma, X., Huang, J., Wei, J., 2016. Toward improved rpl: a congestion avoidance
Conference (CCWC). Las Vegas, NV, USA. multipath routing protocol with time factor for wireless sensor networks. J. Sens.
Pushpalatha, M., Anusha, T., Rao, T.R., Venkataraman, R., L-Rpl, 2021. RPL powered by 1–11.
laplacian energy for stable path selection during. Comput. Network. 184, 107697. F. Teraoka, K. Gogo, K. Mitsuya, R. Shibui and K. Mitani, 2008, Unified Layer 2 (L2)
Qiu, T., Liu, J., Si, W., Wu, D.O., 2019. Robustness optimization scheme with multi- Abstractions for Layer 3 (L3)-Driven Fast Handover, IETF, RFC 5184, IETF,1-29.
population Co-evolution for scale-free wireless sensor networks. IEEE/ACM Trans. Thomson, S., Narten, T., Jinmei, T., 2007. IPv6 Stateless Address Autoconfiguration.
Netw. 27 (3), 1028–1042. IETF, RFC 4862, 1–30.
Rakesh, B., Sultana, H.P., November 2021. A review on enhanced routing solutions in Thomson, C., Romdhani, I., Al-Dubai, A.Y., Qasem, M., Ghaleb, B., Wadhaj, I., 2016.
RPL protocol. Int. J. Perform. Eng. 17 (11), 938–945. Cooja Simulator Manual. Edinburgh Napier University, Edinburgh.
Redi, J., Basu, P., 2004. Effect of overhearing transmissions on energy efficiency in dense Thu, N.Q., Nien, L.M., Khoi, N.N., 2015. Multipath RPL protocols for greenhouse
sensor networks. In: IPSN ’04 Proceedings of the 3rd International Symposium on environment monitoring system based on Internet of Things. In: IEEE 2015 12th
Information Processing in Sensor Networks. Berkeley, California, USA. International Conference on Electrical Engineering/Electronics, Computer,
Safaei, B., Mohammadsalehi, A., Khoosani, K.T., Zarbaf, S., Monazzah, A.M.H., Samie, F., Telecommunications and Information Technology (ECTI-CON). Hua Hin, Cha-am,
Bauer, L., Henkel, J., Ejlali, A., 2020. Impacts of mobility models on RPL-based Thailand.
mobile IoT infrastructures: an evaluative comparison and survey. IEEE Access 8, Thubert, P., 2012. Objective Function Zero for the Routing Protocol for Low-Power and
167779–167829. Lossy Networks (RPL), IETF. RFC 6552, 1–13.
Safaei, B., Monazzah, A., Ejlali, A., 2021. ELITE: an elaborated cross-layer RPL objective Triantafyllou, A., Sarigiannidis, P., Lagkas, T.D., 2018. Network protocols, schemes, and
function to achieve energy efficiency in internet of things devices. IEEE Internet mechanisms for internet of things (IoT): features, open challenges, and trends.
Things J. 8 (2), 1169–1182. Wireless Commun. Mobile Comput. 1–24.
Saleem, A., Afzal, M.K., Ateeq, M., Kim, S.W., Bin Zikria, Y., 2020. Intelligent learning Tsvetkov, T., 2011. RPL: IPv6 Routing Protocol for Low Power and Lossy Networks.
automata-based objective function in RPL for IoT. Sustain. Cities Soc. 59, 102234. Network Architectures and Services, pp. 59–66.
Salman, T., Jain, R., 2017. Networking protocols and standards for internet of things. In: Tyan, H., Gao, Y., Hou, J., 2019. Tutorial: Working with J-Sim. . [Online]. Available: htt
Internet of Things and Data Analytics Handbook. John Wiley & Sons, Inc. ps://sites.google.com/site/jsimofficial/j-sim-tutorial
Sampayo, S.L., Montavont, J., Noël, T., 2019. LoBaPS: load balancing parent selection for Varga, A., 2001. The omnet++ discrete event simulation system. In: The European
RPL using wake-up radios. In: IEEE Symposium on Computers and Communications Simulation Multiconference. Budapest, Hungary.
(ISCC). Barcelona, Spain. Vasseur, J.-P., Dunkels, A., 2010. Interconnecting Smart Objects with IP: the Next
Sebastian, A., Sivagurunathan, S., 2018. Multi DODAGs in RPL for reliable smart city IoT. Internet. Morgan Kaufmann.
Journal of Cyber Security and Mobility 7 (1), 69–86. Vaziri, B.B., Haghighat, A.T., 2020. Brad-OF: an enhanced energy-aware method for
Sehgal, A., Mayzaud, A., Badonnel, R., Chrisment, I., Schönwälder, J., 2014. Addressing parent. Wireless Pers. Commun. 114 (5).
DODAG inconsistency attacks in RPL networks. In: Global Information Infrastructure Vera-Pérez, J., Todolí-Ferrandis, D., Santonja-Climent, S., 2018. A joining procedure and
and Networking Symposium (GIIS). QC, Canada, Montreal. synchronization for TSCH-RPL wireless sensor networks. Sensors 18 (10), 3556.
Sennan, S., Ramasubbareddy, S., Nayyar, A., Nam, Y., Abouhawwash, M., 2021. LOA- Verma, A., Ranga, V., 2020a. Security of rpl based 6LoWPAN networks in the internet of
RPL: novel energy-efficient routing protocol for the internet of things using lion things: a review. IEEE Sensor. J. 20 (11), 5666–5690.
optimization algorithm to maximize network lifetime. Comput. Mater. Continua Verma, A., Ranga, V., 2020b. Mitigation of DIS flooding attacks in RPL-based 6LoWPAN
(CMC) 69 (1), 351–371. networks. Transactions on Emerging Telecommunications Technologies 31 (2),
Sesli, E., Hacıoğlu, G., 2017. Contiki OS usage in wireless sensor networks (WSNs). e3802.
Turkish Journal of Electromechanics & Energy 2 (2), 1–6. Voigt, T., 2012. Contiki cooja crash course. In: Intl. School on Cooperative Robots and
Sethi, P., Sarangi, S.R., 2017. Internet of things: architectures, protocols, and Sensor Networks.
applications. Journal of Electrical and Computer Engineering 1–25. Wang, J., Chalhoub, G., 2019. Mobility support enhancement for RPL with multiple
Seyfollahi, A., Ghaffari, A., 2021. A review of intrusion detection systems in RPL routing sinks. annals of telecommunications 74 (1).
protocol based on machine learning for internet of things applications. Wireless Watteyne, T., 2014. OpenWSN. . [Online]. Available: https://openwsn.atlassian.net
Commun. Mobile Comput. 2021, 1–32. /wiki/spaces/OW/overview?homepageId=688187
Seyfollahi, A., Moodi, M., Ghaffari, A., 2022. MFO-RPL: a secure RPL-based routing Watteyne, T., Vilajosana, X., Kerkez, B., Chraim, F., Weekly, K., Wang, Q., Glaser, S.,
protocol utilizing moth-flame optimizer for the IoT applications. Comput. Stand. Pister, K., 2012. OpenWSN: a standards-based low-power wireless development
Interfac. 82, 103622. environment. Transactions on Emerging Telecommunications Technologies 23 (5),
Shafique, U., Khan, A., Rehman, Abdur, Bashir, F., Alam, M., 2018. Detection of rank 480–493.
attack in routing protocol for low power and lossy networks. Annals of Winter, T., Thubert, P., Brandt, A., Hui, J., Kelsey, R., Levis, P., Pister, K.P., Struik, R.,
Telecommunications 73 (15), 429–438. Vasseur, J., Alexander, R., 2012. RPL: IPv6 routing protocol for low-power and lossy
Sharma, D., Shukla, A.N., 2014. A comparative study of the routing protocols LOAD and networks. IETF, RFC 6550, 1–157.
RPL in low-power and lossy networks (LLN). Scholars J. Eng. Technol. 2 (1), 85–87. Witwit, A., Idrees, A.K., 2018. A comprehensive review for RPL routing protocol in low
Shreenivas, D., Raza, S., Voigt, T., 2017. Intrusion detection in the RPL-connected power and lossy networks. In: New Trends in Information and Communications
6LoWPAN networks. In: SIA CCS ’17: ACM Asia Conference on Computer and Technology Applications. Springer International Publishing, pp. 52–66.
Communications Security. Abu Dhabi United Arab Emirates. Yang, Z., Liu, H., Chen, Y., Zhu, X., Ning, Y., Zhu, W., 2021. UEE-RPL: a UAV-based
Sneha, K., Prasad, B.G., 2017. Load balanced efficient routing technique for mobility energy efficient routing for internet of things. IEEE Transactions on Green
based RPL. In: Second International Conference, SmartCom. Pune, India. Communications 5 (3), 1333–1344.
Sobhan babu, B., Lakshmi Padmaja, P., Ramanjaneyulu, T., Lakshmi Narayana, I., Zanella, A., Bui, N., Castellani, A., Vangelista, L., Zorzi, M., 2014. Internet of things for
Srikanth, K., 2017. Role of COOJA simulator in IoT. International Journal of smart cities. IEEE Internet Things J. 1 (1), 22–32.
Emerging Trends & Technology in Computer Science (IJETTCS) 6 (2), 139–143. Zhang, K., Liang, X., Lu, R., She, X., 2014. Sybil attacks and their defenses in the internet
Sobral, J.V.V., Rodrigues, J.J.P.C., Rabêlo, R.A.L., Al-Muhtadi, J., Korotaev, V., 2019. of things. IEEE Internet Things J. 1 (5), 372–383.
Routing protocols for low power and lossy networks in internet of things Zhang, L., Feng, G., Qin, S., 2015. Intrusion detection system for RPL from routing choice
applications. Sensors 19 (9), 2144. intrusion. In: IEEE International Conference on Communication Workshop. ICCW),
Solapure, S.S., Kenchannavar, H.H., 2020. Design and analysis of RPL objective functions London, UK.
using variant routing metrics for IoT applications. Wireless Network 26, 4637–4656. Zhao, M., Ho, I., Chong, P.H.J., 2016. An energy-efficient region-based RPL routing
Solapure, S.S., Kenchannavar, H.H., Sarode, K.P., 2020. Issues faced during RPL protocol protocol for low-power and lossy networks. IEEE Internet Things J. 3 (6),
analysis in contiki-2.7. In: ICT Systems and Sustainability. Springer, Singapore, 1319–1333.
pp. 477–485. Zhao, M., Kumar, A., Chong, P.H.J., Lu, R., September 2017. A comprehensive study of
Song, C., Liu, M., Cao, J., Zheng, Y., Gong, H., Chen, G., 2009. Maximizing network RPL and P2P-RPL routing protocols: implementation, challenges and opportunities.
lifetime based on transmission range adjustment in wireless sensor networks. Peer-to-Peer Networking and Applications 10 (5), 1232–1256.
Comput. Commun. 32 (11), 1316–1325.

52
K.A. Darabkh et al. Journal of Network and Computer Applications 207 (2022) 103476

Muna Al-Akhras received the B.Sc. degree in computer engi­

neering from Jordan University for Science and Technology,
Irbid, Jordan, in 2011, and the M.Sc degree in Computer and
Khalid A. Darabkh received the B.Sc. degree in Computer Networks Engineering from University of Joradn, Amman,
Engineering from Yarmouk University in 2002 with honors. Jordan,in January 2021 She is currently working as a Research
During the training term, which was a BSc curriculum Assistant at the University of Jordan. Her research interests
requirement at Yarmouk University, he was honored by both include wireless networks, sensor networks, and Internet of
the Computer and Communication Systems Co. and Yarmouk Things.
University as of being the first trainee who not only got
employed during the training term, but also started training its
technical staff (engineers and technicians). He received the M.
S.E. degree in Electrical and Computer Engineering from New
York Institute of Technology, USA, in 2003 with distinction.
Due to obtaining the University of Jordan Scholarship, he
received the PhD degree in Computer Engineering from the
University of Alabama in Huntsville (UAH), USA, in 2007 with
honors. He was selected for inclusion in the Who’s Who Among Students in American
Universities and Colleges. He was honored by UAH for being not only so outstanding, but Jumana N. Zomot received the B.Sc. and M.Sc. degrees in
also finishing the PhD requirements along with publishing a massive number of reputable Computer Engineering from University of Applied Science
scientific articles in record time. He has joined the Computer Engineering Department at Jordan, in Sep. 2005 and the University of Jordan, Amman,
the University of Jordan as an Assistant Professor since 2007 and obtained a distinguished Jordan, in May 2019. She is currently working at the Infor­
professorship in 2016. He authored and co-authored of at least a hundred ninety highly mation Technology Directorate, Ministry of Transport,
esteemed research articles. He is among World’s Top 2% Scientists List, for various dis­ Amman, Jordan. Her research interests include wireless sen­
ciplines, compiled by Stanford University in 2020 and 2021. Prof. Darabkh is the recipient sors networks and computer networks.
of 2020 Federation of Arab Scientific Research Councils Reward – Theme of Innovation.
Prof. Darabkh is further the recipient of Ali Mango Distinguished Researcher Reward for
Scientific Colleges and Research Centers in Jordan. He is among Top 5 JU Researcher’s List
who published the highest number of quality manuscripts in Scopus. Not only to this
extent, but rather he is Sun Certified Programmer for Java Platform (SCJP), Oracle
Database Administrator Certified Associate (OCA), as well as Oracle Java Developer
Certified Trainer (OJDC-T). In fact, he serves on the Editorial Board of Telecommunication
Systems, published by Springer, Computer Applications in Engineering Education, pub­
lished by John Wiley & Sons, and Journal of High Speed Networks, published by IOS Press.
Additionally, he serves as a TPC member of highly reputable IEEE conferences such as Mohammed Atiquzzaman ([email protected]) [M′ 87- SM′ 95]
GLOBECOM, ICC, LCN, VTC-Fall, PIMRC, ISWCS, ATC, ICT, and IAEAC. Moreover, he is a holds the Edith Kinney Gaylord professorship of Computer
member of many professional and honorary societies, including Eta Kappa Nu, Tau Beta Pi, Science at The University of Oklahoma, USA. He serves as the
Phi Kappa Phi, and Sigma XI. He was listed among those of IBC Leading Scientists of the Editor-in-Chief of Journal of Network and Computer Applica­
World, Cambridge, England as well as IBC 2000 Outstanding Scientists, Cambridge, En­ tions, Founding Editor-in-Chief of the Vehicular Communica­
gland. He was selected for inclusion in the Marquis Who’s Who in the World. He super­ tions journal, associate editors of IEEE Communications
vised and examined at least 100 masters’ theses in most of Jordanian universities. As Magazine, IEEE Transactions on Mobile Computing, etc. He
administrative experience at the University of Jordan, he served as Assistant Dean for received the IEEE 2018 Satellite and Space Communications
Computer Affairs in the College of Engineering from Sept 2008 to Sept 2010. Additionally, Technical Recognition Award for valuable contributions to the
he served as Acting Head of the Computer Engineering Department from June 2010 to Sept Satellite and Space Communications scientific community. He
2012. At the end, he is engaged in research mainly on Internet of things, Software-defined also received the IEEE 2017 Distinguished Technical Achieve­
networks, vehicular networks, flying ad-hoc networks, Fog networking, Full duplex ment Award in recognition of outstanding technical contribu­
cognitive radio networks, queuing systems and networks, multimedia transmission, tions and services in the area of communications switching and
channel coding, steganography and watermarking, as well as innovative and interactive routing. In recognition of his contribution to NASA research, he received the NASA Group
learning environments. Achievement Award for “outstanding work to further NASA Glenn Research Center’s effort
in the area of Advanced Communications/Air Traffic Management’s Fiber Optic Signal
Distribution for Aeronautical Communications” project. His research interests and publi­
cations are in next generation computer networks, wireless and mobile networks,
switching and routing, optical communications and multimedia over networks.

53