International Journal of Intelligent Networks 5 (2024) 204–211

Contents lists available at ScienceDirect

International Journal of Intelligent Networks

journal homepage: www.keaipublishing.com/en/journals/
international-journal-of-intelligent-networks

Adaptive switching and routing protocol design and optimization in

internet of things based on probabilistic models
Yi Yang
Teaching and Research Support Center, Naval Submarine Academy, Qingdao, Shandong, 266000, China

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

Keywords: Through smart cities, Intelligent Transportation Systems (ITS), the agricultural sector, and wearable devices, the
Internet of things Internet of Things (IoT) has revolutionized several human interests. Through the development of new cluster
Optimization tasks, the Decision-Making System (DMS) of Cluster Heads (CHs), and improving the accuracy of traffic pre­
Fuzzy Logic
diction and reliability of transportation, the present study intends to improve the energy depletion of IoT devices.
Cuckoo Search
Objective function
The paper explores the subject of data flow optimization using Fuzzy Assisted Cuckoo Search Optimization
Routing (FACSO), traffic flow using Gaussian Process Regression (GPR), and CH prediction using the Stochastic Opti­
Energy consumption mization Algorithm (SOA). Optimizing network lifetime while minimizing Energy Consumption (EC) is feasible
through the practical application of the SOA, GPR, and FACSO models. Increasing End-to-End Delay (EED),
Network Throughput (NT), and energy efficiency can be rendered feasible through a real-time DMS regarding
routing employing a novel approach referred to as FACSO. This approach has enhanced the efficacy and reli­
ability of Wireless Sensor Networks (WSN). With up to 500 nodes and an EC of 0.3451 J, the experiment’s
findings demonstrate that a proposed SOA-FACSO model achieves superior EED.

1. Introduction become a reality. The reconfiguration will ensue by making physical

devices’ smart,’ allowing them to accomplish things independently,
The Internet of Things (IoT) has got attention in recent years. It is a resulting in the emergence of the IoT. By connecting anywhere at any
potential technology that can provide remedies to challenges that arise time but in any location, IoT promises to make advanced devices more
in multiple sectors. The Internet serves as a network’s communication available.
backbone for exchanging information from one platform to another IoT primarily allows for the autonomous yet safe communication of
globally. The IoT concept came into existence in 1998, and in 1999, real-world objects. IoT reduces physical work by automating regular
Kevin Ashton coined the term IoT at the Massachusetts Institute of processes. The number of Internet-connected objects is continually
Technology (United States) Auto-ID lab [1]. It’s an impenetrable but expanding. Smartphones contain a range of sensors that detect data,
intelligent system that detects, regulates, and can be reprogrammed compute it, and send the relevant data over the Internet. Using that
through embedded systems to communicate with one another. With system with diverse devices and sensors, several applications can be
high productivity improvements, IoT allows instant access to informa­ made to provide substantial benefits. IoT intelligent things can be
tion related to any device. According to the CISCO team, around 50 uniquely identified. These gadgets use Radio-Frequency Identification
billion smart devices have already been connected to the Web [2]. (RFID) tags or barcodes, which are detected by sensor devices [3].
A billion connected devices comprise the IoT, which can detect, The data acquired by the sensors is sent to the processing system over
collect, and send data from one device to another without human sup­ the Internet for analysis. The processed findings are forwarded to the
port. Health care tracking, building automation, logistics, linked autos, Decision-Making System (DMS) and Action-Invoking System (AIS),
smart city development, smart grid, smart home, smart retail, smart which performs the appropriate action. Important information and fa­
agriculture, and other IoT-enabled services all contribute to amelio­ cilities are constantly available, making it easy to design new apps,
rating people’s lives. The Internet will become obsolete in this scenario, introduce new work, connection, and performing methods, and
and a new idea of connected smart’ things will emerge. IoT can trans­ contribute to a new lifestyle. As a result, there will be much traffic, and
form the Internet so that Machine-To-Machine (M2M) learning will humans will be in the minority of traffic generators and recipients [4].

E-mail address: [email protected].

https://doi.org/10.1016/j.ijin.2024.05.001
Received 28 March 2024; Received in revised form 30 April 2024; Accepted 1 May 2024
Available online 8 May 2024
2666-6030/© 2024 The Author. Published by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Y. Yang International Journal of Intelligent Networks 5 (2024) 204–211

Because of the issues and opportunities that IoT brings, it is being (b) To boost the link accuracy in WSN by developing the dynamic
researched in numerous research areas. In this instance, maintaining a and EC effective parents’ decision in RPL.
consistent model is required to store all the data efficiently. (c) To determine efficient software-defined networking-based WSN
Routing is the method of sending data from one node to another. The using Machine Learning (ML) to enhance QoS in WSN.
routing protocol’s performance is accomplished by lowering the route
cost. Efficient route selection saves energy during packet delivery and Select the optimum parent according to the parent’s predicted en­
extends the Network’s Lifetime (NL) [5]. Route discovery and mainte­ ergy to try out the RSS system.
nance protocols provide information about the journey from the source The remainder of the article is organized as follows: Section 1: the
node (SrN) to the destination (DsN). The node is usually in one of two overview of IoT, WSN, and routing protocols are detailed in the Intro­
modes: storing or non-storing. The participant node collects and trans­ duction; Section 2: the related work, comprehensive analysis, research
mits data to its parent in the storage mode. In non-storing mode, the gap, motivation, and objective of the research are given in Related
participant node conveys data to its parent node. Proactive, reactive, Works, Section 3: the proposed Adaptive switching and routing protocol
and hybrid protocols are route discovery and maintenance [6]. based on probabilistic model is given in Proposed methodology, Section
The routing procedure is accepted based on the node’s location in a 4: the simulation analysis with illustration is shown in Result and Dis­
network. Network structure-based routing enhances the Quality of cussion, and Section 5: the article is concluded with recommendation for
Service (QoS) while lowering network node Energy Consumption (EC). future is presented in Conclusion.
Hierarchical, flat, and location-based routing are the three network
models. Routing functions such as delivery mechanisms and computa­ 2. Related Works
tion methods are used to operate the system [7]. Wireless Sensor Net­
works (WSN) link the virtual and physical worlds—tiny sensors coupled For the routing process, the Routing Protocol for Low-Power and
to one another transfer values to the Internet. Sensors are implemented Lossy Networks (RPL) protocol was enhanced as a well-known resolu­
in the network to monitor many physical and environmental features in tion in its Machine-to-machine (M2M) connection because of RPL’s
a WSN. This network comprises many nodes, each of which has a precise flexibility. The RPL routing’s qualifications have been prepared
particular function. Using Internet Protocol (IP)-based WSN technolo­ to meet several requirements of dissimilar Low Power and Lossy Net­
gies, the WSN can be connected to the internet [8]. works (LLN) applications. The principal uniqueness of the RPL protocol
WSNs are made up of nodes that can be used in several applications, can be estimated in this research work, where the crucial purpose is to
including health care, agricultural, and military applications [9]. WSNs provide the reader with complete information about the RPL protocol. In
are used to monitor many applications over the internet. It permits RPL, numerous problems need to be located for the ever-present utili­
physical objects linked together through nodes, activators, and con­ zation of applications of the LLNs in the physical surroundings [13].
nections to collect data. Several academics proposed WSN systems built The optimal route amongst two Sensor Nodes (SN) in the LLNs
for real-world applications based on different requirements. It describes cannot be provided by default since the buffer’s size and memory are
how WSN is a backbone for IoT by considering an example of smart cities limited. Consequently, path discovery will be required to sustain the
[10]. Connecting sensors, lights, and meters are the IoT devices that Point-to-Point (P2P) communication. In a reactive routing protocol, the
smart cities employ to collect and evaluate data. After obtaining this optimal route has been discovered by applying the P2P-RPL. This has led
data, smart cities can use it to enhance and improve their public services, to a considerable overhead in large and dense networks [14]. In the
infrastructure, and amenities. An intelligent, secure environment is network, all the SNs in WSN need to cooperate in structuring the pro­
within approach with the support of these intelligent things. visional Destination Oriented Directed Acyclic Graph (DODAG) during
A division of the IoT that will end up in improved transport networks the pathfinding [15]. However, this is directed to enhance EC. In LLNs, it
is called the Internet of Vehicles (IoV). In IoV, there are three commu­ is used to discover the optimal route that considers EC. In the protocol
nication possibilities: Vehicles-to-Vehicles (V2V), Vehicles-To- structure, the reliability must be measured [16].
Roadsides (V2R), and Roadsides-To-Roadsides (R2R). VANETs are uti­ For universal data traffic, LLN purposes required fixed routing sup­
lized in several applications, including vehicle speed monitoring, traffic portiveness. The P2P, Point-to-Multipoint (P2MP), and Multipoint-to-
conjunction avoidance, optimal route finding, and communication be­ Point (MP2P) with heterogeneous facilities are obtainable for the SN
tween outside and inside cars [11]. As a result, WSNs are a collection of in WSN. The RPL has constructed the DODAGs in behaviour that can be
specialized objects that provide sensing services to IoT devices. When it used in the routing to optimize MP2P data traffic. However, the P2P and
involves detecting, monitoring, and tracking different environmental P2MP’s broadcasted information must move on to the pre-designed DAG
tasks, WSNs are essential networks. The rapid development of commu­ method. Consequently, RPL has not provided the effectual routing, and
nication and networking systems has resulted in a demand for the the message delivery goes through from high EED and lossy wireless
planning, building, and executing of multiple intelligent (unmanned) links. The P2P communication required by frequent LLN appliances
service networks [12]. must be maintained by a mediator agent in the process of RPL [17].
In optimizing and designing adaptive switching and routing pro­ The DODAG root, in the manner of a non-storing procedure, has
tocols for the IoT, probabilistic models play a prominent role. By denoted an intermediate SN. As a result, it can be restricted access with a
examining patterns of network traffic and device connections, these high range of data traffic. In the DODAG method, the P2MP data traffic
models permit effective routing decisions based on dynamic environ­ is sustained by utilizing downward paths. Besides, the downward routes
ments. A case study could emphasise employing a protocol in a smart may not be optimal since the wireless channels contain asymmetric
city development, where IoT devices connect for numerous applications behaviour. Thus, this type of asymmetric character of wireless links has
like environmental monitoring, traffic management, and public services. affected the routing algorithms’ performance in LLNs. While they have
Through continuous adjustment and monitoring of routing approaches been utilized for asymmetric links, the protocols do not carry the
utilizing probabilistic models, the network can adapt to altering condi­ asymmetry of links that can provide fail-in real-time networks [18].
tions, optimize resource utilization, and ensure reliable communication The traffic conjunction has taken place in LLN for both the connec­
within the IoT ecosystem. tion and levels of an SN in a WSN. In the case of more traffic, the
inadequate node’s buffer size can be directed to the overflow process
1.1. Contribution of the research [19]. Besides, the traffic conjunction will be directed to reducing
channel quality and PLR extension. Thus, this will lead to more EEDs,
(a) Using a successful routing design with RPL, this study intends to struggle on the channel, and packet drop, which will also be enhanced.
improve link accuracy, EC, and QoS in WSNs. Currently, many techniques are being developed to overcome traffic

205
Y. Yang International Journal of Intelligent Networks 5 (2024) 204–211

conjunction problems, such as [20]. 4.1. Network formulation

For LLNs, two significant objectives must be used to enhance routing
protocols: energy-saving and reliable data or message delivery. How­ The dynamic WSN is expressed via an undirected graph, denoted by
ever, multicast routing is maintained by the RPL rouging, and it can the G=(L, SN). In this graph, the accumulation of links among the SN is
select one SN as a superior parent exploiting definite object tasks. In the denoted as ‘L,’ and the set of SN is denoted as SN. The link between the
lossy wireless channels and dynamic changing topology, this single SN ‘y’ and ‘x’ in graph G, where the nodes rely on the communication
route routing cannot be sufficiently measured [21]. In wireless channels, range.
packets’ data, packet rebroadcasting, and PLR have been enhanced The core concept of WSN is defined as follows.
during this unreliability process. This will be directed to a high level of
EC and an extended time of channel possession. (a) Nodes residing in the network pose unique node identity (ID),
The challenges of RPL have been discussed under severe conditions, and before the incidence of fusion, there was no awareness about
including the discovery of energy-efficient routes, effectual routine the network other than its ID. Entire node information will be
maintenance for general traffic patterns with inadequate memory, removed once it leaves the network region, and the node ID will
Jamming finding and controlling, Energy-efficient and Reliable routing be for further transmission.
in LLNs, High percentage of throughput with a low duty cycling of lower (b) In the WSN, the Base Station (BS) is available in the preceding
layer, mobility support, issues in security method, complexity of RPL, link, and the setup of BS has already been attained.
under specification of RPL protocol, trickle junction and loops. (c) The distinct node can replicate the response or broadcast, which
accomplishes a local estimation and is fixed interval and coor­
3. Research gap dinated time. Every period, a single node is activated as a receiver
or transmitter, and in some instances, both will act together. The
The research gap lies in addressing the limitations and challenges of data in the sink or gathering gaps are designated as round.
the RPL protocol in LLN. Despite its flexibility, RPL faces suboptimal (d) Collusion identification is not available in the network.
routing, energy inefficiency, and susceptibility to jamming. Addition­ (e) The nodes in the adjacent position can communicate, and the
ally, there’s a need for more efficient multicast routing and consider­ information among the nodes is symmetric. It depicts a specific
ation of dynamic network conditions. Addressing these gaps could node, and ‘y’ permits the data from the ‘x’ node. The node ‘x’ also
enhance routing reliability and energy efficiency in LLN. allows the message from ‘y’.

Nodes in the graph are segregated into several groups that are stated
3.1. Motivation as clusters, and these clusters accomplish G’s sub-graph. In the formu­
lated cluster, a distinct node that acts as a center for the sub-graph, the
Modern routing algorithms employed by LLN present multiple errors CH, exists and is linked with all the nodes. The cluster members generate
and issues, which is why this research is needed. After a thorough re­ a form of network topology.
view, it has been concluded that these techniques require significant
changes in order to provide accurate and effective routing. So, to
enhance WSN-QoS, EC, and total accuracy, novel and better routing 4.2. Cluster task
methods are essential. Route and data distribution efficiency is the
highest priority today due to the increasing importance of IoT In WSN, the CH selection technique is attained by the fitness value
technologies. utilized to elect CHs from the SN. The Fitness Function (FF) encompasses
New methods for cluster tasks, Cluster Head (CH) selection, traffic two significant measures: the usage of energy by the CH and nodes in the
simulation, and routing efficiency are the main objectives of this study, network. Secondly, total EC combines the information at the CH point
which attempts to address this vital requirement. Using cutting-edge and propagates the message to the BS. The EC between the CHs and SN is
optimization techniques, including Fuzzy Assisted Cuckoo Search valued by EQU (1) to EQU (4).
Optimization (FACSO), Gaussian Process Regression (GPR), and Sto­ j ∑{ }
∑ Egi − wfitness fun (ji )
chastic Optimization Algorithms (SOA), we can improve cluster tasks for Egi = × ji (1)
Egmaxi − Egmini
optimal NL or reduce EC by decreasing the impact on unpredictability in j=1 ∀nj ∈j
i

node features and the state of the network. Node features such as battery
level, processing capability, and communication range influence task wfitnessfun (ji ) = op lvl(i) + WT + Dc + Rm (2)
allocation and routing decisions. Network state variables like traffic load [ ]
and congestion levels guide efficient data distribution and minimize Egresidual energy ( )
energy wastage. Investigation significantly improves End-to-End Delay WT = WTmaxi 1− × Avg vij (3)
Eginitial energy
(EED), Network Throughput (NT), and EC in WSNs by implementing
new routing optimization methods like FACSO, which update real-time (
ni
)
routing decisions based on WSN circumstances. The intention is to Dc = < trange (4)
distance(i, j)
address the challenge of practical and unpredictable connectivity in IoT-
enabled settings by addressing the drawbacks of present-day routing where the level of operator is indicated as op_lvl, the weight of the node
technologies. is indicated as WT, and the relative mobility is indicated as ‘Rm’ that
occurs among the node and CH, and the degree of connectivity is indi­
4. Proposed model-adaptive switching and routing protocol cated as ‘Dc’ where all the nodes in the network form the cluster.
based on probabilistic models (ASRP-PM) Stochastic optimization-based routing utilizes stochastic optimiza­
tion techniques to challenge the uncertainties and variabilities present
The proposed technique, ASRP-PM, is accomplished via a clustering in network environments. Initially, the problem of routing is identified
scheme that devices to assign EC between the clusters at a stable time, by employing this approach, implementing the advantage of unpre­
thus developing NL. The primary intent of ASRP-PM is to minimize the dictability like link quality, traffic flow, and failures of nodes. Then,
EC across the sensor device, reduce the EED, and reduce message adopting a model of stochastic challenges, targets like optimizing NT or
overhead for the admittance of real-time data. The NL of WSN is minimizing EED are determined. EQU (5) defines the Objective Function
exploited. (OF) for maximizing the routing efficiency, F(x).

206
Y. Yang International Journal of Intelligent Networks 5 (2024) 204–211

∑
f(x) = xij
μ (5) Table 1
(i,j)∈E ij
Fuzzy variable range.

where the expected EED is given as (i, j). Node Features Fuzzy Variable Input Range
It defines the problem in the context of stochastic variables, such as Node Degree Low Less Than 1
the energy of nodes and traffic flows. OF aims to improve parameters Medium 1 to 4
such as NL or EC within stochastic constraints. Selections are driven by High Greater Than 4
Node Energy Low 0.0–0.35
stochastic designs, which describe limitations. In order to effectively Medium 0.36–0.65
address the issue, SOA techniques such as Genetic Algorithms (GA) are High 0.66–1
used. To ensure adaptability, adaptive processes change CH selection Node ED Low Less Than 1 ms
following real-time data. The technique has been enhanced via testing Medium 1 ms–3 ms
High More than 3 ms
and parameter setting. Applying this method, we may improve reli­
ability and performance in dynamic network environments through the
development of CH methods of selection that make the most of sto­ 4.4. Fuzzy Assisted Cuckoo Search Optimization-based routing
chastic optimization. In Algorithm 1, we can observe the phases of the
recommended method. This proposed FACSO approach is implemented in the controller
Algorithm 1 for ASRP-PM node. The parameter node degree, EED, and energy are collected from
Step 1. Function Cluster Task WSN the nodes and passed as the input to the fuzzy system. The node with
Step 2. For Each SN ‘i’ in WSN high energy participates in the Cuckoo Search (CS) algorithm to find the
Step 3. Compute Eg_i using EQU (1) optimal path from the SrN to DsN.
Step 4. Compute w_FF (j, i) using EQU (2)
Step 5. Compute WT using EQU (3)
The FL system assumes input as node degree, energy, and EED and
Step 6. Compute Dc using EQU (4) produces the output, whether the node is intruder or regular.
Step 7. For Each SN ‘i’ in WSN: This fuzzy system works as a three-step process.
Step 8. For Each neighbouring node ‘j’:
Step 9. If Eg_i-w_FF (j_i) ≥ Eg_Maxi - Eg_Mini
• Fuzzification: In this step, the complex input of node degree, en­
Step 10. Assign node ‘j’ to cluster of node ‘i’
Step 11. Update cluster task probabilities based on traffic models ergy, and EED is converted into a fuzzy set.
Step 12. Apply SOA-based routing to optimize cluster tasks • Inference: Based on the provided rules, it applies the fuzzy input set
Step 13. Return optimized clustered WSN and produces the fuzzy output.
Step 14. End • Defuzzification: In this step, the predicted output is converted to a
complex value, and the node is determined as either an intrusion or a
normal node.

4.3. Probabilistic traffic model using GPR The following are the linguistic and fuzzy input variables needed to
find intrusion nodes.
Regression and probability classification problems are solved using
the general supervised ML technique, GPR. For small data sets, it • Node Degree – (Low, Medium, High)
accurately forecasts the uncertainty measures. By giving each function a • Node Energy Level – (Low, Medium, High)
probability value and calculating the mean of the probability distribu­ • Node EED- (Less, Normal, More)
tion among the functions, it can identify the function that best fits the
data out of many tasks while accurately describing the facts. The pre­ The input range of each fuzzy variable is given in Table 1. The lin­
dictable regression function has a degree of confidence in this proba­ guistic variables used for the output variable node type are Rather Risk,
bilistic technique. Risk, High Risk, Low Risk, Moderate, Low Safe, High Safe, Safe, and
The next value is computed using data flow, and the prior function Rather Safe.
space is stated to compute the predictive subsequent distribution. As in Rule 1: If the node degree is > 1, then the node degree is Low,
multivariate Linear Regression (LR), in GPR, time is also reserved, all indicating a sparse network connectivity.
nodes in the path, initial and final packets are measured as independent Rule 2: If the node degree is between 1 and 4, then the node degree is
variables, and status is a dependent variable. Medium, suggesting moderate network connectivity.
By giving the function y = θx + e, the probability function p(θ) is Rule 3: If the node degree is < 4, then the node degree is High,
computed. P(θ) represents the previous distribution on the parameter indicating dense network connectivity.
‘θ’. The subsequent distribution is computed using the Bayes rule given Rule 4: If the node energy is between 0.0 and 0.35, then the node
in EQU (6). energy is Low, implying low energy reserves.
( )
( ) P y, θ p(θ) Rule 5: If the node energy is between 0.36 and 0.65, then the node
θ
P ,x =
x
() (6) energy is Medium, suggesting moderate energy reserves in the node.
y p xy Rule 6: If the node energy is between 0.66 and 1, then the node
energy is High, indicating ample energy reserves in the node.
( ) ( ) Rule 7: If the node EED is less than 1 ms, then the node EED is Low,
where P is the subsequent distribution, P xy, θ is the probability
y, x
θ
indicating minimal EED in data transmission.
() Rule 8: If the node EED is between 1 and 3 ms, then the node EED is
and prediction of xy being true if the given θ is true p xy , and is the Medium, suggesting moderate EED in data transmission.
marginal probability. Rule 9: If the node EED is more than 3 ms, then the node EED is High,
Then, this model’s GPR of prediction of the ‘y’ value is assumed in indicating significant EED in data transmission.
EQU (7). The CS algorithm is a meta-heuristic algorithm that uses the FF to
∫ ( ) approximate the fitness level of a node using the parameters link NL,
θ
p(D, y, x) = p(D, θ)p , x dθ (7) bandwidth, and node energy. The FF is used for steadier routes, thus
y
θ improving the Packet Received Rate (PRR). The FF used in this approach
is the primary function, the sum of energy, NL, and bandwidth param
where ‘y’ is the prediction label, and ‘x’ is the test observation.

207
Y. Yang International Journal of Intelligent Networks 5 (2024) 204–211

Fig. 1. Flowchart of FACSO

eters calculated. The calculation of FF is given in EQU (8).

∑ Energy = Initial Energy − Utilized Energy (11)
(Energy + bandwidth + lifetime)
f(xi ) = (8)
Distance(xi − xi− 1 ) In order to identify the optimal relay node, all three of these variables
are employed to determine the CS fitness. Fig. 1 demonstrates the flow
The parameters link NL, energy, and bandwidth used in the FF chart of the FACSO approach. Each network node’s parameters, such as
calculation are computed as follows:
degree, EED, and energy, are initially inputted into the FL system to
The time required for two nodes to communicate over a route is identify high-energy nodes. A node with high energy will participate in
known as the Link Lifetime (LL). In order to identify a significant secure
the CS algorithm. Levy flight chooses a random channel and detects the
path, a factor analysis is performed. Data Packet Loss Rate (PLR) can be neighbourhoods of the current node ‘i’.
decreased with the support of this secure route direction. It is well-
Each neighbourhood node is evaluated using the FF. The best one is
defined as the variance between the distance Δd over time Δt, EQU (9). selected as the next hop node among the neighbours. Similarly, from the
Δd new hop, the process repeats to choose the next hop in the path. This
LL = 1 − (9) process continues until it reaches the final destination node.
Δt
The procedure of the FACSO is assumed in the Algorithm 2.
The lower probability of link failure and higher communication
Algorithm 2. Procedure of FACSO
effectiveness of data packets define efficient LL connecting two nodes.
Bandwidth analysis is used to avoid data PLR in WSN. The study also Step 1. Start FACSO
Step 2. Get Node Degree, Energy, EED
resulted in minimizing traffic. Consider that CB refers to bandwidth
Step 3. Apply Fuzzy Logic System
capacity and SQ to backlog size. EQU (10) provides the Bandwidth Step 4. If (Node has the highest EED, highest energy, and lesser node
Availability (BA). connectivity)
Step 5. Node with higher energy
SQ Step 6. Else
BA = 1 − (10)
CB Step 7. Normal node
Step 8. End Fuzzy logic system
Each node in the network was first provided with an appropriate Step 9. Launches CS
level of energy—the sending and receiving of packets needed a partic­ Step 10. Measure its fitness Fl
ular level of energy from the nodes. EQU (11) provides the process for Step 11. If (Fk < Fl)
computing the node’s residual energy level. Step 12. Update l as a new solution
(continued on next page)

208
Y. Yang International Journal of Intelligent Networks 5 (2024) 204–211

Table 2 Table 5
Comparison of NT (bytes). Comparison of EC (J).
No of Node RPL LLN FACSO No of Node RPL LLN FACSO

100 2904 2504 8404 100 0.797 0.767 0.456

200 4358 4138 16,495 200 0.834 0.782 0.493
300 5451 5101 21,264 300 0.867 0.845 0.534
400 5997 5787 24,659 400 0.898 0.856 0.552
500 6895 6295 27,177 500 0.899 0.887 0.576

Table 3
Comparison of EED (ms).
No of Node RPL LLN FACSO

100 0.199922 0.181922 0.141717

200 0.691281 0.491201 0.294001
300 0.965479 0.800479 0.446285
400 1.56976 1.10976 0.598570
500 1.83556 1.40357 0.743240

Table 4
Comparison of PRR (count).
No of Node RPL LLN FACSO

100 153 150 151

200 281 284 264
300 289 300 330
400 313 346 378
500 387 400 413

Fig. 2. Comparison of NT
(continued )
Algorithm 2. Procedure of FACSO
transmitting a packet from the initial node to its destination of choice is
Step 13. Continue the best solutions indicated by time EED. PRR is determined as the ratio of the total
Step 14. End CS packets reached through the target node from the complete packets sent.
Step 15. Transmit the Data through the best solution
The PRR is the ratio of the number of packets delivered to the target
Step 16. Stop FACSO
Step 17. End If node to the total number of packets sent. The performance of the algo­
Step 18. End rithms are given in Table 2 to Table 5. The performance is measured
using EQU (12) to EQU (15) (see Table 3).
Total No. of Bytes transmitted
NT = (12)
Time Taken
5. Result and Discussion
EED = Time at packet received − Time at packet transmitted (13)
The FACSO improved routing efficiency and was implemented using
No. of Packet Reached
the NS-2 simulator with 500 nodes. Any source that requests to transmit PRR = (14)
No. of Packet Sent
the data requests the controller node for the entire route to the desti­
nation. The controller uses the CS algorithm to find the optimal path
Etx = Ptx × Ttx (15)
from the source to the destination. The simulation setup encompasses
different parameters defining the features and conditions of the simu­ where the EC during data transmission is given as Etx , duration is Ttx and
lated WSN environment. It employs a wireless channel with a 2-way energy is Ptx .
ground radio-propagation model, Wireless PHY network interface
type, and 802_11 MAC type for communication. Packet queuing is 6. Discussion
managed by a PriQueue, with LL as the link layer type and Omni_An­
tenna as the antenna model. The comparative analysis of the routing protocols, RPL, LLN, and
The interface queue is configured to accommodate a maximum of 50 FACSO, across several performance metrics reveals distinct performance
packets. The simulation involves 500 mobile nodes employing the differences. FACSO consistently outperforms RPL and LLN in terms of
6LoWPAN routing protocol to facilitate communication. The topog­ NT, with significantly higher NT values recorded across all numbers of
raphy of the simulated network is defined with dimensions of 900 by nodes: 8404, 16,495, 21,264, 24,659, and 27,177 bytes for 100, 200,
614 units. The simulation duration is set to 20.0 units of time. These 300, 400, and 500 nodes, respectively, compared to RPL and LLN.
parameters collectively establish the framework for conducting simu­ However, FACSO exhibits higher EED values compared to RPL and LLN,
lations to evaluate the performance and behaviour of wireless networks with delay values ranging from 0.141717 ms to 0.743240 ms for FACSO
under specified conditions and scenarios. The measures throughput, across different numbers of nodes, while RPL records the lowest EED
PRR, EC, and EED are used to show the performance of the IoT end nodes values ranging from 0.199922 to 1.83556 ms.
using the FACSO for optimal route finding against RPL and LLN (see Additionally, FACSO demonstrates lower EC than RPL and LLN, with
Table 4). energy consumption values of 0.456–0.576 J for FACSO, whereas RPL
NT is a rate that measures the number of data sent within a particular and LLN show relatively higher EC, ranging from 0.767 to 0.899 J.
time in the network’s bandwidth. The total amount of time used for

209
Y. Yang International Journal of Intelligent Networks 5 (2024) 204–211

Fig. 3. Comparison of EED

the total amount of nodes changes significantly. Yet FACSO consistently

displays lower EC than RPL and LLN, indicating it could be the better EC
approach.
Fig. 4 illustrates this finding, suggesting that FACSO frequently
shows lower EC than RPL and LLN. In general, when comparing NT,
EED, and EC, FACSO performs more effectively than RPL and LLN,
suggesting that it could be an improved routing protocol. Testing the
results and investigating the hidden variables resulting in performance
variations involves more investigation and testing (see Fig. 5).
Once validated through simulations, the adaptive protocols are
deployed in a controlled environment within the company’s facilities.
Real-time monitoring tools track key performance metrics such as PDR,
EED, and NT.

7. Conclusion and future work

The Internet of Things has brought given about transformative al­

terations across diverse domains, namely Intelligent Transportation
Systems, smart cities, wearable devices, and agriculture. This research
work focuses on improving energy efficiency in Internet of Things de­
Fig. 4. Comparison of PRR vices by enhancing traffic prediction, data flow, and decision-making
system of cluster heads. With the assistance of the innovative applica­
Despite these differences, all three protocols exhibit similar trends in tions of Fuzzy Assisted Cuckoo Search Optimization, Gaussian Process
PRR, with minor variations observed across different numbers of nodes. Regression, and Stochastic Optimization Algorithm, the research pro­
These numerical comparisons underscore FACSO’s potential as a more longs network lifetime while minimizing energy consumption. The real-
efficient routing protocol, particularly in scenarios prioritizing high NT time decision-making system for routing, utilizing Fuzzy Assisted
and energy efficiency. Cuckoo Search Optimization, enhances network throughput, end-to-end
The NT, EED, PRR, and EC analysis reveals distinct performance delay, and energy efficiency but also improves the consistency of wire­
features among the RPL, LLN, and FACSO routing protocols. In contrast less sensor networks. This novel technique offers promising prospects for
to LLN and FACSO, RPL consistently shows lower NT across numerous optimizing Internet of Things device performance and network consis­
nodes, with FACSO consistently achieving the greatest NT. This struc­ tency in diverse applications.
ture is depicted visually in Fig. 2, where the highest NT is identified in Future innovations may include using artificial intelligence and
FACSO, followed by LLN and RPL. Compared to LLN and FACSO, RPL optimized performance methods motivated by nature.
has the most rapid data transmission due to its minimal EED ratios.
However, the highest EED ratios can be detected in FACSO, which might Funding acknowledgements
suggest that data transmission EED is not practical.
The variances are illustrated visually in Fig. 3, where the lowest Not Applicable.
delay can be seen with RPL next to LLN and FACSO. As demonstrated by
Fig. 3, PRR tends to be constant within RPL, LLN, and FACSO, even when

210
Y. Yang International Journal of Intelligent Networks 5 (2024) 204–211

Fig. 5. Comparison of EC

Data Availability [8] S. Sankar, S. Ramasubbareddy, A.K. Luhach, A. Nayyar, B. Qureshi, CT-RPL: cluster
tree-based routing protocol to maximize the lifetime of Internet of Things, Sensors
20 (20) (2020) 5858.
Not Applicable. [9] M. Mahyoub, A.S. Hasan Mahmoud, M. Abu-Amara, T.R. Sheltami, An efficient
RPL-based mechanism for node-to-node communications in IoT, IEEE Internet
CRediT authorship contribution statement Things J. 8 (9) (2021) 7152–7169, https://doi.org/10.1109/JIOT.2020.3038696.
[10] D. Pancaroglu, S. Sen, Load balancing for RPL-based internet of things: a review,
Ad Hoc Netw. 116 (2021) 102491.
Yi Yang: Writing – review & editing, Writing – original draft, Vali­ [11] M. Mahyoub, A.S.H. Mahmoud, M. Abu-Amara, T.R. Sheltami, An efficient RPL-
dation, Supervision, Software, Methodology, Investigation, Funding based mechanism for node-to-node communications in IoT, IEEE Internet Things J.
8 (9) (2020) 7152–7169.
acquisition, Formal analysis, Conceptualization. [12] A.K. Idrees, A.J. Witwit, Energy-efficient load-balanced RPL routing protocol for
Internet of Things networks, Int. J. Internet Technol. Secur. Trans. 11 (3) (2021)
Declaration of competing interest 286–306.
[13] A. Dalvin Vinoth Kumar, L. Arockiam, Route discovery overhead aware routing
protocol for WSN to enhance QoS, Int. J. Control Theor. Appl. 9 (27) (2016)
The authors declare that they have no known competing financial 249–254.
interests or personal relationships that could have appeared to influence [14] J. Yao, C. Yan, J. Wang, C. Jiang, Stable QoE-Aware Multi-SFCs cooperative
routing mechanism based on Deep Reinforcement learning, IEEE Transactions on
the work reported in this paper. Network and Service Management 21 (1) (2024) 120–131, https://doi.org/
10.1109/TNSM.2023.3287601.
References [15] K. Haseeb, N. Islam, A. Almogren, I. Ud Din, H.N. Almajed, N. Guizani, Secret
Sharing-based energy-Aware and Multi-hop routing protocol for IoT based WSNs,
IEEE Access 7 (2019) 79980–79988.
[1] K.A. Darabkh, M. Al-Akhras, J.N. Zomot, M. Atiquzzaman, RPL routing protocol
[16] S. Rout, B. Monna, F. Pareschi, G. Setti, W.A. Serdijn, Spread-spectrum Modulated
over IoT: a comprehensive survey, recent advances, insights, bibliometric analysis,
Multi-channel Biosignal acquisition using a Shared analog CMOS Front-end, IEEE
recommendations, and future directions, J. Netw. Comput. Appl. 207 (2022)
Transactions on Biomedical Circuits and Systems 17 (4) (2023) 872–884, https://
103476.
doi.org/10.1109/TBCAS.2023.3317188.
[2] Z.A. Almusaylim, A. Alhumam, N.Z. Jhanjhi, Proposing a secure RPL based internet
[17] Hanane Lamaazi, Nabil Benamar, Antonio J. Jara, RPL-based networks in static
of things routing protocol: a review, Ad Hoc Netw. 101 (2020) 102096.
and mobile environment: a performance assessment analysis, Journal of King Saud
[3] N. Zahedy, B. Barekatain, A.A. Quintana, RI-RPL: a new high-quality RPL-based
University - Computer and Information Sciences 30 (3) (2018) 320–333.
routing protocol using Q-learning algorithm, J. Supercomput. 80 (2024)
[18] S. Vastrad, K.R. Shobha, Power-efficient trickle algorithm with dynamic
7691–7749, https://doi.org/10.1007/s11227-023-05724-z.
adjustment of Idle interval for internet-of-things application, IEEE Sensor. J. 24 (3)
[4] S. Moradi, R. Javidan, A new objective function for RPL routing protocol in IoT to
(2024) 3853–3862, https://doi.org/10.1109/JSEN.2023.3341426.
increase network lifetime, Int. J. Wireless Mobile Comput. 19 (1) (2020) 73–79.
[19] W. Zhang, S. Zhu, J. Tang, N. Xiong, A novel trust management scheme based on
[5] S. Cherbal, A. Zier, S. Hebal, et al., Security in Internet of things: a review on
Dempster–Shafer evidence theory for malicious nodes detection in wireless sensor
approaches based on blockchain, machine learning, cryptography, and quantum
networks, J. Supercomput. 74 (4) (2018) 1779–1801.
computing, J. Supercomput. 80 (2024) 3738–3816, https://doi.org/10.1007/
[20] A. Musaddiq, R. Ali, S.W. Kim, D.-S. Kim, Learning-based resource management for
s11227-023-05616-2.
low-Power and lossy IoT networks, IEEE Internet Things J. 9 (17) (2022)
[6] Z. Yang, H. Liu, Y. Chen, X. Zhu, Y. Ning, W. Zhu, UEE-RPL: a UAV-based energy-
16006–16016, https://doi.org/10.1109/JIOT.2022.3152929.
efficient routing for Internet of Things, IEEE Transactions on Green
[21] J. Gao, J. Fan, S. Zhai, G. Dai, Message passing based wireless Multipath SLAM
Communications and Networking 5 (3) (2021) 1333–1344.
with continuous Measurements Correction, IEEE Trans. Signal Process. 72 (2024)
[7] A. Agiollo, M. Conti, P. Kaliyar, T.N. Lin, L. Pajola, DETONAR: detection of routing
1691–1705, https://doi.org/10.1109/TSP.2024.3376083.
attacks in RPL-based IoT, IEEE transactions on network and service management
18 (2) (2021) 1178–1190.

211