This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS 1

A Trust-Aware and Authentication-Based

Collaborative Method for Resource Management
of Cloud-Edge Computing in Social
Internet of Things
Alireza Souri , Senior Member, IEEE, Yanlei Zhao , Mingliang Gao , Asghar Mohammadian , Jin Shen ,
and Eyhab Al-Masri , Senior Member, IEEE

Abstract— The Social Internet of Things (S-IoT) paradigm I. I NTRODUCTION

is focused on topic of the Internet of Things (IoT), which
accelerates the object issues by working with the concept of
social networks. Searching and finding a new object in the
community are considered to manage the number of friends
T HE Internet of Things (IoT) is a network of billions of
heterogeneous and often intelligent devices with Internet
access that continuously receive a continuous stream of data
and complex relationships between them and affect the ability to from the environment or generate a continuous stream of infor-
navigate at the cloud-edge layer, and resources, such as battery
lifetime of S-IoT devices and energy resources, are important mation so that interconnected objects are available anywhere in
challenges in this field. In the processing of social messages the world and at any time and interact with each other or with
of remote devices, increasing the battery life of devices that humans [1]. The Social IoT (S-IoT) is an evolutionary model
require such requirements plays the most important role. In this of IoT, in which objects range from smart things to intelligent
research, a collaboration scenario is presented to consider object objects for collaborating with other smart devices and social
attributes, friend’s functions and intelligent friend selection
among objects for group messaging. First, a general reference applications to generate more accurate and diverse data to
model is designed and presented to select a friend to access provide better cloud-edge services [2]. Using the principles
group message remote processing services and minimize cloud- and rules of human social networks and applying them to
edge resources. The simulation results show that, for the correct the IoT have the following advantages for the development of
communication of friends at the edge of the network and in the edge network capabilities of the IoT in terms of resource
each service discovery, according to the length of the path in
the network, it is possible to establish stable communication and access.
make better service with the least possible. The results show that 1) Limiting devices from human social manners in small
if we want to develop a method for friendship between objects communities and with a more limited application of
in communication in cloud computing, the proposed method can
S-IoT infrastructure rules in relation to different peers
greatly improve the effectiveness of providing reliable message
processing types. provides more collaborative capabilities in computing
and processing at the edge of the IoT network [3].
Index Terms— Group message processing, Internet of Things
(IoT), resource management, social environment, trust-based
2) Scalability in S-IoT shows the role of developing and
authentication, trust management. accessing objects in managing their social relationship
at the level of social relationship of human society,
i.e., discovering mutual services and carefully selecting
Manuscript received 25 October 2022; revised 23 December 2022 and friends by consulting network members and evaluating
15 January 2023; accepted 24 January 2023. (Corresponding author:
Yanlei Zhao.) social trust conditions [4].
Alireza Souri is with the School of Electrical and Electronic Engineering, 3) The cooperation of elements in S-IoT regardless of
Shandong University of Technology, Zibo 255000, China, and also with the their platforms, and the creation of social links between
Department of Software Engineering, Faculty of Engineering, Haliç Uni-
versity, 34060 Istanbul, Turkey (e-mail: [email protected]; alirezasouri@ objects enables the interaction of heterogeneous devices
halic.edu.tr). at the edge of the network [5].
Yanlei Zhao, Mingliang Gao, and Jin Shen are with the School of
Electrical and Electronic Engineering, Shandong University of Technology, A smart social network must have a link of common interests,
Zibo 255000, China (e-mail: [email protected]; [email protected]; feelings, and experiences to maintain social relationships with
[email protected]). mutual trust and confidence. These social relationships ensure
Asghar Mohammadian is with the Department of Computer Engineer-
ing, Ilkhchi Branch, Islamic Azad University, Ilkhchi 5358114418, Iran a reliable and intelligent communication space for colleagues
(e-mail: [email protected]). and stakeholders to acquire knowledge, develop information,
Eyhab Al-Masri is with the School of Engineering and Technol- and exchange ideas in the network [6]. In addition, back-
ogy, University of Washington Tacoma, Tacoma, WA 98402 USA
(e-mail: [email protected]). ground knowledge of qualitative, operational, and quantitative
Digital Object Identifier 10.1109/TCSS.2023.3241020 characteristics of objects, by imitating human social behavior,
2329-924X © 2023 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS

increases the expansion of relationships between objects in consumption to meet quality requirements in this approach.
the IoT network. The influential criteria in the social relations The rest of this article is organized as follows: The background
of devices at the edge of the IoT network are their past of the S-IoT scenario is reviewed in Section II. In Section III,
interaction, the choice of data processing location, distance, a general model that analyzes S-IoT is presented and the net-
and energy consumption of the devices in the communication work relationships in this model are formulated to evaluate the
system [7]. In the communication of objects, the accuracy system performance, in Section IV. The proposed algorithm
of event detection and the scalability of the system and its for selecting a processor friend is shown in Section V and the
availability while maintaining trust in its performance are operational simulation and presentation of the results are also
important [8]. However, the extent and diversity of devices shown in Section VI. In Section VII; future objectives and
and the area of communication search at the edge of the final points have been discussed.
network are growing rapidly, which, along with the speed of
data growth and the dynamic diversity of the S-IoT environ- II. R ELATED W ORK
ment, has caused more complexity in the interactive model S-IoT is a dense pattern with multiple objects that enable the
of the devices, and it is the complexity of battery energy interaction of heterogeneous elements with multiple features
consumption, which is of great importance. This issue makes and various services in the community of objects. These
it necessary to choose the appropriate task processing position elements include people, objects, or devices that communicate
in the IoT social network. To face the search for a suitable task with communication technologies such as Bluetooth, Wi-Fi,
processing device at the edge of the network, it is important and LoRa. The devices of this social network of things can: 1)
to consider the independent information processing model of interact their interaction automatically or stop communicating
each device, the reliability of the service providers, and the and create a personal network community; 2) process and exe-
interaction history between the edge devices in the network. cute the desired executive programs and services by exploring
For this reason, objects must have the capacity to process the network; and 3) announce their presence together with
data and tasks, the capacity to store data, and the energy other elements of the S-IoT paradigm to provide services [9].
profile of their neighbor friends. The most important challenge Identification of elements in the network includes the behavior
in the social communication of elements at the cloud-edge of objects in terms of location, specifications, and required
layer is the number of neighboring friends, which affects the services provided by them and influenced by various aspects
amount of memory usage, processing energy consumption, of each object such as ownership, throughput, communica-
and service discovery efficiency. Of course, the number of tion bandwidth, processing, and dynamics. The abundance of
friends in each applicable area is related to the communication interaction in the IoT network and especially at the network
scenario. In this research, optimal processing location selection edge facilitates the discovery of common peers at the network
for improving system performance by properly naming among edge. In [10], a new method for designing IoT platforms is
the available devices is presented to process a task and also used as architecture-independent infrastructure that can design
enables network navigation. systems without any knowledge of programming languages or
The existing key contributions of the proposed method are electronics. Chen et al. [8] investigated the trust factor and
shown as follows: criteria such as community benefits, honesty, and cooperation
in social links between IoT devices and their communication
1) designing a communication model of interconnected
with each other dynamically. Interactions between devices are
objects through the exchange of resources and services
updated based on the distributed trust protocol in a communi-
in the IoT social network;
cation session. They also showed a tradeoff in communication
2) detection and possible analysis of social relationships
convergence rate for processing fluctuations in a low-trust
between objects at the edge of the IoT by describing
environment. The authors have investigated the detailed plan-
their characteristics;
ning for convergence, flexibility, and operational integrity of
3) investigating the dynamic and static aspects of commu-
the communication network, simultaneously with the issue of
nication between objects with an interaction-based view
trust in a changing environment. In the multiagent environment
as well as a distance-based view;
of the IoT, despite the existence of multiple agents consisting
4) proposing the optimal algorithm for choosing the right
of different working groups, the characteristic of social rela-
friend in edge layers;
tions between objects and neighbor discovery, the principles
5) computing of the loss function of this algorithm to
of safety and trust assigned by the owners [3], a reliable link
search for position and estimate its parameters while
for social cooperation in service delivery. It makes it possible
obtaining optimization results for the use of network
for a new member to benefit from the entire network at any
processing resources.
time and this is also effective in improving the quality of
In this processing overhead, the limited elements and end services (QoS) and the speed of task processing and better
devices are offloaded to the central processing resource, and resource management [11]. Fortino et al. [12] introduced a
the smart friend selection strategy is used according to the CoT scenario in which physical objects in the cloud were
characteristics of introducing elements in the network to connected with software agents and virtual to use the social
eliminate the tracking of irrelevant aspects. The use of decision network method for cooperation among themselves. They
theory to optimize the selection and optimal friend selection at divided the existing tasks into various groups with character-
the edge of the network helps for decreasing human resource istics of cooperation and ambition. When configuring a group

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

SOURI et al.: TRUST-AWARE AND AUTHENTICATION-BASED COLLABORATIVE METHOD FOR RESOURCE MANAGEMENT 3

of objects, they benefit from accepting the local reputation

of elements in the social network of objects by avoiding
heavy computation and processing and by being aware of the
system through cooperation with their community agents. For
evaluating community of objects, social capital is proposed
in [13] to maximize the benefit of individual elements in the
network and the entire community in services. The result is
an objective function based on relationships between agents.
The operation is recommended. They have used a mecha-
nism of competition with rewards that ensures meritocracy in
operational execution.
In other research work, Hong et al. [14] used a game theory
mechanism for evaluation of existing QoS-based computa-
tional offloading in an IoT system with cooperation between
network edge devices. They have investigated the problem Fig. 1. Sample S-IoT interactive model for IoT devices.
of multihop computing load in IoT edge cloud computing
model during intercommunity routing of IoT devices. Also,
according to [15], the interaction of processing tasks, data, virtual and physical layers change as data operations, but
and resources between a group of end devices of the IoT at the primary information is generated through the interaction
the edge of the network has a positive and improving effect between objects or by the user, and the second-level informa-
on energy consumption. Alatoun et al. [16] contributed a new tion is continuously generated through the operation of virtual
energy-aware Task Scheduling method for real-time Internet of entities. After the processing service is requested by the IoT
Medical Things. It has proposed a task scheduling algorithm end device, a friend is selected based on the defined rating
to enhance some QoS factors. It has also proposed a special and reliability. Then, service delivery is initiated through the
weight for each node that is, computed on the estimated energy communication infrastructure and implemented to improve the
consumption values. Simulations were performed in iFogSim2 utilization of processing, energy, and storage resources [17].
to evaluate the performance of the proposed framework. The However, the solution [18] has drawbacks that include the
proposed method schedules tasks efficiently with optimal QoS following.
factors. 1) Communication management and selection of the
appropriate position of the device (appropriate friend)
III. G ENERAL M ODEL OF S-I OT IN P ROVIDING in the processing connection is only in the S-IoT
C LOUD -E DGE S ERVICES server, and if the connection fails, the network is
disconnected.
In the S-IoT paradigm, objects are given the characteristics 2) This model is suitable according to the problem demand
of providing services and searching for information similar of network edge communications that are short range.
to human performance in groups. In this paradigm, objects 3) This system does not include gateway objects because
are allowed to create links between themselves to create joint communication management is disabled in the S-IoT
functionality and imitate human social behavior in searching physical layer.
and retrieving and transferring information, requesting ser- 4) The overhead problem is caused by handing over pro-
vices. S-IoT enables intelligent interactions between objects cessing and data from devices with limited resources,
instead of enhancing intelligence in them, and this enables such as bandwidth, energy, and processing limitations,
the socialization of objects [15]. In the S-IoT infrastructure, which itself creates limitations in transferring data to the
reliable data must be collected, classified, and analyzed in each S-IoT server.
domain [9]. We first consider a three-layer model architecture There are some problems during the process such as the need
for S-IoT at the network edge as follows: 1) the physical layer for local cooperation, the need for navigation in the edge areas
consisting of IoT devices; 2) the virtual layer that provides of IoT, the need for continuous communication to transfer data
the IoT social network; and 3) the application layer. The related to owner cooperation activities to create a common
lowest layer physically creates a dynamic environment by place in a local area, and target social relations. According to
“things” and delivers processing data and assigned tasks to the above problems, discovery and search remain of the service
the upper layer. The second-level communication contracts of partner’s best friend is provided in our model. Fig. 1 shows the
social virtual objects (SVOs) in this layer physical objects sample S-IoT interactive model for IoT devices relationship.
are connected with social capabilities. They are responsible
for processing the data and tasks received from the network
elements in the lower layer, and while analyzing the data, they IV. T RUST-BASED C OMMUNICATION T RENDS
provide the appropriate response for the existing user requests. IN S-I OT N ETWORK
Then, the cloud-edge layer is responsible for providing the Object relations between the social networks in the IoT are
desired services. With the implementation of the system, the considered to the intellectual-cognitive capacity of humans for

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS

participating the performance of each object in its life cycle Therefore, in order for nodes to reach the cooperating objects
and each period of membership in the IoT network. Com- with high accessibility and ease, it is necessary to determine
municating some inherent characteristics of objects during a the efficient diameters in each connection that connects all the
communication affects mutual cooperation in each relation- objects of a region. To estimate the characteristics of each
ship, such as object type selection, mobility model selection, object in the S-IoT model and the type of its relationship
operating model type, device processing power, and commu- with its peers, we define a probability distribution with a
nication frequency in each interaction. Allocating, tracking, continuous random variable X which depends on the distance
and retrieving these relationships between objects in S-IoT are parameters, the type and the working group of the requested
done without human intervention and only with his permission service, two connected objects in any type of relationship
at the application level. Focusing on the cooperation of things CLOR, CWOR, SOR, POR, and OOR keeps the social contin-
with each other and with humans in S-IoT, the following uously connected [20], [21]. Considering the features related
relationships are formed autonomously, which provide services to the interaction history and target workspace, because of the
such as processing and use of other network resources in the involvement of nodes in routing and positioning in the IoT
appropriate situation [4]. network and because of the participation of more nodes in
more than one correlation of service provision, the importance
1) Parent-Object Relationship (POR): This relationship is
of keeping records is visible. By carefully examining the
established between homogeneous devices and is fixed
probability density function of a uniform distribution for a
and consistent at almost any time based on device pro-
random variable assigned to the relationship and the type of
file parameters such as product category, manufacturer,
expected service defined by X [20], any type of relationship
category, and end device class.
between objects can be specified, analyzed, and defined [22].
2) Co-Location Communication of IoT Objects Together
(C-LOR): Between homogeneous/heterogeneous devices
located at a fixed location, this communication is formed A. Friendship Relationship Between Objects Based on
solely based on the device’s position in the IoT network Probabilistic Trust
hierarchy. Bayesian deduction has been used to correctly predict the
3) Relationship of IoT Objects Based on the Type of friendship relationship between objects based on interaction
Task and Common Work They Perform (CWOR): This and evaluate the trust probability of dynamic friendship rela-
relationship is formed based on providing a common tionship. Considering fixed features such as the node builder
task between homogeneous/heterogeneous devices. This in the simulator environment, the interaction duration of
relationship is established to implement the decision neighboring objects is calculated randomly. Therefore, because
at the required time and is almost constant in every of the dynamic interaction changes with time, a dynamic
operating period of the system. friendship with a uniform distribution between objects can be
4) Relationship Based on the Object of Ownership (OOR): defined for this interaction and Bayes theorem can be used to
This concept is provided for heterogeneous devices with model it as follows:
a common owner.
PDF(Friend|I )
5) Social Object Relationship (SOR): This relationship is
PDF(I | Friend) ∗ PDF(X | Friend) ∗ PDF(Friend)
shown a collaborative method between heterogeneous =
cloud-edge devices and based on frequent previous con- PDF(I ) ∗ PDF(X )
(1)
nections between objects that belong to related owners
(such as devices belonging to classmates, friends, and where I is the amount of interaction and X is the appearance
different social groups). distance between friends in the local friendship relationship.
Device selection based on the type of communication and
requested services can be determined by all three parameters, B. Decisioning Offloading Tasks and Processes to Friend
the number of steps of the shortest route, geographical dis- and Peer Devices
tance, and social location [9]. The possible distance distri- The goal of this section is to determine end-device process-
bution of network nodes in an area is based on the number ing for existing task offloading procedures for a high-level
of steps of the shortest path of each pair of nodes and is processing partner device, which tasks can be processed on
randomly selected. Finally, the position of the nodes in each a network end device or a peer device. By measuring the
connection in the social network of objects is effective on the performance value of the system during the time and energy
number of selected nodes that have a friendship and social consumption of the process discharge in the edge device of
relationship and are looking for a particular service. In the the IoT, the issue of maximum and optimal use of processing
developed communication model, there is no restriction on resources, communication, and optimal allocation of resources
object selection to receive services such as task processing, can be raised at the edge.
and all nodes have an equal chance to start a friendship In exchange for energy consumption, the IoT network can
communication in S-IoT [19]. According to the mentioned use the resource management approach to ensure QoS at the
S-IoT model, each device on the edge of IoT can commu- edge, reduce latency, and ultimately save energy to solve this
nicate with other objects and devices based on trust-based problem. Considering the magnitude and continuity of the data
authentication and its history, location, and profile parameters. flow collected from/to the environment by end devices at the

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

SOURI et al.: TRUST-AWARE AND AUTHENTICATION-BASED COLLABORATIVE METHOD FOR RESOURCE MANAGEMENT 5

end of the IoT edge, the altruism and interactive relationship SOR is the distribution parameter of objective social relations
between them described and the selection of the processing between objects and CWOR is a shared task relation between
transfer location, which is explained in the operational model objects. Each of the functions of relation (1) is the cumulative
next, it is possible to check the different performance condi- density of relations for the better performance of objects in
tions of the processing services of the devices at the edge of establishing relations. Thus, various types of friendship rela-
the network. Sometimes it is even necessary to perform the tionships, static friendship, and dynamic friendship between
initial processing of tasks at the edge of the IoT in the device pairs of nodes have been used.
itself, which is called self-cooperation, which depends on the The network distance between the end device and the
processing task. gateway device does not change over time and device profile
In S-IoT, considering that the phenomenon of friendship data that are always constant. Dynamic Friendship is one that
can be formed as a gradual relationship in the community of changes over time, such as history of interaction of process-
objects at the edge of the network, therefore, the selection ing tasks between different levels of devices at the network
of friends between objects is done based on the principles edge. Based on this, the type of interactive communication
of homophily [23]. For example, the more similar two edge to provide processing services at the edge of the network is
devices are to each other in terms of features, the more defined based on distance and processing and data assignment.
likely they are to be friends [23] and thus be in the first In fact, each node in any type of intranetwork communication
level of cooperation in edge processing. These friendships and at the edge of the network updates its assessment of others
collaborations can be defined as C-LOR or CWOR at the IoT while communicating and based on the type of relationship or
edge level. According to the type of relationship of friendship interactive behavior between them.
and cooperation between objects and the density of uniform Therefore, according to the above explanations in the type
distribution between them, X Friend can be considered with all of relationships between network edge devices, three methods
the independent variables of the relationship defined between for processing at the edge of the IoT are suggested as follows:
objects. Therefore, with the uniformity of the probability of
1) direct connection between IoT end devices and edge
availability of XPOR and XCLOR, all objects involved in a
servers, to process tasks in edge servers according to
relationship at the edge of the IoT network and subtracting
S-IoT collaboration relationship definition;
from one by removing cooperation cases that do not have
2) local processing at IoT end devices or at local gateways
any of the defined relationships between objects, at least one
at the edge of the network (node-peer friendly relation-
variable survival relation between objects in the local region
ship);
can be calculated as follows and considered in the process
3) local execution of tasks on the edge servers with a
allocation model:
regional scale.
F(friend(x)) = 1 − (1 − FPOR(x)) ∗ (1 − FCLOR(x)) The ability to perform processing on the network edge device
∗(1 − FOOR(x)) ∗ (1 − FSOR(x)) and upload tasks to the edge for processing depends on
∗(1 − FCWOR(x)). (2) the capabilities of the network edge devices and the com-
prehensive definitions of the cooperation model such as tit-
By calculating the cumulative distribution function and proba- for-tat mechanism in S-IoT [25]. On the other hand, some-
bility density of X Friend as a uniform distribution, it is possible times excessive and unnecessary delays are caused in the
to show the most probable pairs of nodes in communicating processing of tasks by the cloud-edge data centers. Then, it
friendship and cooperation. According to (2), the X Friend is recommended to perform processing on local network edge
distribution between each pair of objects at the edge of the devices such as IoT edge gateways, which structurally and
network is similar to a power law and guarantees the type of dynamically change the communication and friendship rules
relationship between objects and devices [24]. X itself is a between nodes. Therefore, in edge processing, according to the
design variable based on the type of connection and the dis- area covered and the type of communication between different
tance and friendship probability distribution between objects. devices in terms of network distance and type of ownership
Each of the subfunctions FPOR(x), FCLOR(x), FOOR(x), and device level, a balance can be established between device
FSOR(x), and FCWOR(x) according to the uniformity of energy consumption, task processing, and delay.
the probability of availability of each pair of friendly objects 1) Proposed Trust-Based Relationship Model: The opera-
in a certain distance and the relevant variables is defined tional components of the proposed scenario, shown in Fig. 1,
as follows; consist of a local gateway connected to the edge server, which
(POR) parent relationship of objects that objects have this provides the main processing and computing resources for
property without dependence on position, have the property of the end devices at the edge. Existing IoT devices can be
making long links between objects and with uniform probabil- mobile, which have limited processing resources and energy
ity distribution. The function FPOR(x) is the density value of due to their size, application, and so on. The computing
the regular distribution for the class of objects with long links. capacity of C.Ti is defined for each device i. The input of this
The same thing is true about other relationships; As CLOR model includes Di , program code size(bits), system settings,
is also a common location relation of objects. OOR is the program codes, and the number of processor units to execute
property relationship of their objects and social relations are existing tasks (Ci ). Based on the mentioned parameters and
formed based on it. This parameter has a uniform distribution. assumptions, the computing capability of the local gateway

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS

and its processing time is computed as follows. Assuming up a trust-based cooperation algorithm between objects at the
the atomicity of the processing tasks delegated from the end edge of the network (friend selection) to select the best friends
device, the amount of data sent and the total amount of from the objects at the edge of the IoT as follows. First,
processor unit can be found and applied to the current system a comprehensive search method was used to find possible
performance model. Assume that any processing task can be solutions and identify applicants who have the most selection
processed by a processing device at the edge or on an IoT end conditions to provide the desired services (processing services
device: Fil , the local computing capability of device i can be in this research) with cooperation. For this purpose, we first
calculated in terms of training/s [26] determined all the possible combinations of the probability of
cooperation (friendship) between the nodes, regardless of the
Til = Ci /Fil . (3) type and order of selection, completely mathematically and
Of course, the overhead caused by the data obtained in step with the relation C (Nr, k). In this relation, Nr is the number
3 was not considered in modeling. The transmitted data rate of of cooperating nodes, where 1 ≤ k ≤ Nr. Here, to select
each end device can be adjusted from zero to Ri (Pi ). The total the best partner nodes in each type of relationship between
remote processing time for device i is Til ( f i , Pi ) as follows: devices and between network levels, we introduced a loss
function as follows, whose variables are given as follows: set
Til ( f i , Pi ) = Tit (Pi ) + Tie ( f i ) (4) of friends in relationship (F), number of jumps (D), total
number of edge services (S), and priority of trust value (T ) in
so that the transmission time is related to the size of the
the relationship. Decreasing the output value of this function
offloaded data value Di and the data transmission rate Tit (Pi )
leads to the selection of your colleague and friend node, which
and Tie ( f i ), respectively, and is given as follows:
leads to an increase in interest and an increase in profit. The
Di relationship between the coefficients in this relationship is
Tit (Pi ) = . (5)
w.log2 (1 + ai .Pi ) considered as follows: α1 + α2 + α3 + α4 = 1 and 0 < α1,
α2, α3, α4 < 1. In this research, the number of friends and
Increase/decrease bandwidth consumption of the network
jumping have destructive effects on energy consumption and
edge communication channel in data transmission between
bandwidth, whose coefficients are positive and in the form of
the local device Id and the IoT edge processing station is
α1 and α2, to increase the losses. In addition, the number and
ai = (h i /N0 ). In this case, Tie ( f i ) will be given as follows:
type of tasks show the parameters of satisfaction. The greater
Ci and positive these factors are, the greater the overall benefit
Tie ( f i ) = . (6)
fi from the S-IoT approach will be. α3 and α4 are negative
coefficients in the loss and damage reduction function of nodes
According to the task offloading method, the required process-
cooperation in a processing optimization problem. The loss
ing energy management remains in the IoT end device and the
function of the method investigated in this research will be
lifetime of the device is increased [27].
given as follows:

V. T RUST-BASED O PTIMAL F RIEND S ELECTION Loss − function = α1 ∗ number of friends(F)

A LGORITHM + α2 ∗ number of jumps(D)
The trust-based optimal friendship selection algorithm is − α3 ∗ number_of_service(S)
presented to find the best edge partner to perform remote − α4 ∗ trust(T). (7)
processing. Assuming that each node likes the processing time
criteria and the type of friendship relations defined as the In the above relation, the more negative the α3 and α4
requester. If node i meets the expectations of node j in the coefficients are in the above function, the greater the scarcity,
type of processing cooperation relationship according to the and the obtained values will indicate more profit in handing
model mentioned in Section IV-A, it does not necessarily mean over the task to friends. Considering that the loss function
that device j also meets the requirements of i and vice versa. basically shows the difference between the predicted value
Most importantly, in simulation, the number of processor and the actual value in the selection of service partners, in this
friends of partners in establishing relationships is not limited, method, we select a combined list with the lowest loss value
but in practice, due to physical requirements and limitations as the list of the proposed nodes for remote processing coop-
such as communication bandwidth the limitation, the number eration. Therefore, in each operation cycle, new friendships
of elements sending friend requests around and interaction (new partner nodes) are created to provide the desired service
with each other is limited. Therefore, at the beginning of (here remote processing service) by Algorithm 1.
networking to provide processing services at the edge of
the network, requesting nodes are randomly selected. Due to VI. E XPERIMENTAL R ESULTS IN I F OG -S IM
the physical nature of objects in S-IoT, resource limitation at E NVIRONMENT
the edge of the IoT network, and the time-consuming search The iFog-Sim simulator is a programmable tool for model-
to provide a specific service such as processing service, our ing the described approach of our research. Various operations
preferred method is to perform remote processing services. and communication and network environment and communi-
At the same time, service quality requirements are also main- cation elements can be well modeled with this tool. The above
tained. Based on the above description and constraints, we set tool allows the modeler to explore algorithmic behavior and

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

SOURI et al.: TRUST-AWARE AND AUTHENTICATION-BASED COLLABORATIVE METHOD FOR RESOURCE MANAGEMENT 7

Algorithm 1 Friend Selection Algorithm TABLE II

1. Inputs: Primary Friend List C HARACTERISTIC PARAMETERS OF E ACH N ETWORK A RE S HOWN
2. Output: New Friend List
3. Start: {
4. Step1: {For any friend request (Nr ):
5. Calculate: D, S, T, F
6. If Request List Had Prior Interaction With network
7. Get (friend’s List)}
8. Step2: {For K=1 toNr , do
9. Calculate: C(Nr , k)and Loss-Function
10. If Loss-Function Is Minimize Then Available Suggestion List Is New
Friend-List}
11. Step3: {For any composition List of friends
12. Check It Out of Friends List and Prepare Suggested Friends List With Max
C(Nr , k)
13. Replace suggested Friends List as New Friends List}
14. End}

TABLE I
S IMULATION PARAMETERS

Fig. 2. Shortest distance of the network in the path that each member of the
network is able to navigate.

networks in Table II with more or less similar characteristics,

was also examined. From the obtained results, it is clear that
the minimum distance is realized with the proposed S-IoT.
interaction between different devices. It shows the results of
Although the giant network is a fully connected network of
this simulation by means of powerful graphics and appropriate
size O(log n), most independent nodes are accessible from any
data outputs. In this section, different simulation scenarios
edge point.
with key criteria of network structure, technical features, and
Of course, in choosing the shortest network distance to
QoS factors of the existing case studies are performed and
provide processing services, there are also effects that reduce
analyzed according to Table I. Among the simulation parame-
the optimal efficiency of choosing a colleague’s friend. First,
ters, synchronization time, degree and number of nodes, size of
we can manage more complexity to access and receive services
scenario network volume, and so on were considered. Also, the
from near nodes in the proposed IoT network and to achieve
appropriate selection of the environment, the level of element
the expected services (processing), which is caused by the
cooperation at the edge of the network, and the conditions
multiplicity of accesses in the network distances: because the
have significant effects on the results obtained.
smaller the number of node connections, the more nodes are
separated from the network, therefore, the connections are
A. Ability to Navigate and Select Remote Processing separated sooner. A node associated with a larger number of
Elements elements—with a higher degree—is a gateway hub. This is
To describe the friend selection model in providing services because when a node assigns a task for processing, it prefers
at the edge of the S-IoT network, two other networks with to hand it over to a neighbor with more processing power and
almost similar characteristics were produced: the random a higher degree [28]. Because the gateway is more centralized,
network and the giant and very large component of the network paths between nodes receiving remote processing
network [19], which each consider a group of edge ele- services will be shorter.
ments with a limited proportion of the total nodes of the According to Fig. 3, we can observe degree of node distribu-
IoT network in pairs. The specifications and characteristic tion in the suggested S-IoT among other network case studies.
parameters of each network are shown in Table II. Fig. 2 The ability to logically select a friend in each service discovery
shows the shortest distance of the network in the path that based on the connection capacity of the object nodes fortified
each member of the network is able to move reaching the the system network navigation by measuring the average path
desired node, and this remote processing is possible in these value, the grade of distribution factor, and the total number of
paths. Of course, in the remote processing service model, the existing links.
minimum path calculation is considered from the cooperating The degree distribution and capacity of objects to connect
receiving node. For comparison, the hop distance between to the network deeply affect network navigation. A node
arbitrarily selected pairs of nodes, selected nodes in other with a higher degree is called a strongly connected node,

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS

Fig. 3. Node degree distribution.

Fig. 5. Configuration focus based on trust and friendship between the objects
located in the edge region.

This delay indicates the minimum time required to find the

partner nodes to provide processing service in each of the
network models. The exhaustive search to find a partner node
for processing services can be shown as O(n!) [29], [30].
To overcome these inconveniences, using machine learning
methods in further program optimization is very useful to
decreasing time complexity factor.
By adding the number of nodes in the list of colleagues at
the edge of the network to form a group of friends, the initial
delay in the S-IoT network is better. However, the stochastic
method still shows the lowest initial organizing delay in the
Fig. 4. Average latency of deployment configuration at the edge of IoT. remote processing structure because it is possible to make
various choices from every point of the network. As can
be seen in Fig. 5, after a full implementation period of the
proposed solution in simulation mode, the focus configuration
which tends to become a hub. The simulation showed is based on trust and friendship between the objects located
that the logical choice of friends and degree distribution in the area while dispersing in a smaller area and distance
increases the measurability of the number of degrees. In this causes communication with more trust. In the scatter diagram
situation, the superiority of S-IoT compared to other networks for trust-based authentication, the rate of rupture according to
such as random node mode is also shown. Another side of our the trust-based and friendship-based authentication in different
experiment includes the optimization results of friend selection networks is more concentrated in the lower effect distance.
strategy. While in friendship group-based networks, the communica-
The nature of S-IoT effectively supports the connection tion process it happens for message processing with less
relationships of nodes based on their similarity and cooper- delay because fewer steps are taken to converge the method
ation, and with a search strategy to find the best suitable deployment configuration for other models that occur with
conditions for selecting friends to obtain the remote processing greater dispersion. The process of identification and initial
service, the needs of the smallest possible number of friends communication in the network and between objects is the
with the smallest will find. The hop link is assigned to same.
use the lowest possible amount of energy, bandwidth and
other resources in handing over processing. Undoubtedly, the
number of processing tasks assigned from each node affects VII. C ONCLUSION
the complexity of the procedure. In our scheme, the massive This article shows the problem of selecting a partner node
number of processing requests imposes pressure on the partner on friendship trust in the cloud-edge environment to provide
node providing the remote processing service. message processing services at the IoT while having more
Therefore, in order to estimate the qualification of the processing capability and larger energy resources to hand over
applicant according to the conditions of selecting the partner processing to the target node with the S-IoT paradigm. In this
node, we considered the capability described in Section IV, problem, the characteristics of social links in the selection
in each implementation period as a condition and operational of nodes, and the correct types of nodes are considered.
model. The synchronization results of nodes’ average latency Considering the social link patterns in the connection of each
of deployment configuration at the edge of IoT are shown in pair of nodes at the edge of the IoT network, there is an
Fig. 4. opportunity to define a trust-based framework for evaluating

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

SOURI et al.: TRUST-AWARE AND AUTHENTICATION-BASED COLLABORATIVE METHOD FOR RESOURCE MANAGEMENT 9

the cooperation and friendship of objects and a trust frame- [12] G. Fortino, F. Messina, D. Rosaci, and G. M. L. Sarné, “Using trust
work. According to the uniformity of distribution of IoT and local reputation for group formation in the cloud of things,”
Future Gener. Comput. Syst., vol. 89, pp. 804–815, Dec. 2018, doi:
devices in each local area of the edge of the network and the 10.1016/j.future.2018.07.021.
prediction of trust based on the Bayes relationship between [13] G. Fortino, L. Fotia, F. Messina, D. Rosaci, and G. M. L. Sarné,
these objects, the uniform cumulative probability density is “A meritocratic trust-based group formation in an IoT environment
for smart cities,” Future Gener. Comput. Syst., vol. 108, pp. 34–45,
indicative of the total possible friendship connections between Jul. 2020, doi: 10.1016/j.future.2020.02.035.
objects. In this research, a friend selection algorithm is pro- [14] Z. Hong, W. Chen, H. Huang, S. Guo, and Z. Zheng, “Multi-hop coop-
posed to obtain remote services such as processing, which erative computation offloading for industrial IoT-edge-cloud computing
environments,” IEEE Trans. Parallel Distrib. Syst., vol. 30, no. 12,
is based on comprehensive search for orientation and for- pp. 2759–2774, Dec. 2019, doi: 10.1109/TPDS.2019.2926979.
mulation of task processing assignment variables. Finally, [15] F. Samie, L. Bauer, and J. Henkel, “From cloud down to things: An
the target scenario was simulated in the iFog-Sim tool and overview of machine learning in Internet of Things,” IEEE Internet
analyzed with other scale-free case studies of random network Things J., vol. 6, no. 3, pp. 4921–4934, Jun. 2019, doi: 10.1109/JIOT.
2019.2893866.
mode and giant components. The simulation results show the [16] K. Alatoun, K. Matrouk, M. A. Mohammed, J. Nedoma, R. Martinek,
better performance of the S-IoT design at the edge of the and P. Zmij, “A novel low-latency and energy-efficient task scheduling
IoT network in communication navigation and handing over framework for Internet of Medical Things in an edge fog cloud system,”
Sensors, vol. 22, no. 14, p. 5327, Jul. 2022, doi: 10.3390/s22145327.
processing to the remote colleague. In the implementation of [17] L. Wei, J. Wu, C. Long, and B. Li, “On designing context-aware
simulations, the features of the network topology such as the trust model and service delegation for social Internet of Things,”
conditions of the minimum network distance of the nodes IEEE Internet Things J., vol. 8, no. 6, pp. 4775–4787, Oct. 2021, doi:
10.1109/JIOT.2020.3028380.
in the routing process between the growth of the number of [18] L. Atzori, A. Iera, G. Morabito, and M. Nitti, “The social Internet of
links and the minimum number of friends were considered. Things (SIoT)—When social networks meet the Internet of Things: Con-
In future works, this research will create new solutions for cept, architecture and network characterization,” Comput. Netw., vol. 56,
no. 16, pp. 3594–3608, Nov. 2012, doi: 10.1016/j.comnet.2012.07.010.
developing optimal solutions in service delivery management [19] C. Marche, L. Atzori, V. Pilloni, and M. Nitti, “How to exploit the
through service management. Reliability approaches, mobility, social Internet of Things: Query generation model and device profiles’
and QoS requirements in different S-IoT scenarios can be dataset,” Comput. Netw., vol. 174, Jun. 2020, Art. no. 107248, doi:
10.1016/j.comnet.2020.107248.
investigated. [20] V. Mohammadi, A. M. Rahmani, A. Darwesh, and A. Sahafi, “Trust-
based friend selection algorithm for navigability in social Internet of
Things,” Knowl.-Based Syst., vol. 232, Nov. 2021, Art. no. 107479, doi:
R EFERENCES 10.1016/j.knosys.2021.107479.
[21] L. Militano, M. Nitti, L. Atzori, and A. Iera, “Enhancing the navi-
[1] J. Wang, M. K. Lim, C. Wang, and M.-L. Tseng, “The evolution of the
gability in a social network of smart objects: A Shapley-value based
Internet of Things (IoT) over the past 20 years,” Comput. Ind. Eng.,
approach,” Comput. Netw., vol. 103, pp. 1–14, Jul. 2016, doi: 10.1016/j.
vol. 155, May 2021, Art. no. 107174, doi: 10.1016/j.cie.2021.107174.
comnet.2016.03.007.
[2] R. Magdich, H. Jemal, and M. B. Ayed, “A resilient trust manage-
[22] S. V. Akram et al., “Performance analysis of IoT and long-range
ment framework towards trust related attacks in the social Internet
radio-based sensor node and gateway architecture for solid waste
of Things,” Comput. Commun., vol. 191, pp. 92–107, Jul. 2022, doi:
management,” Sensors, vol. 21, no. 8, p. 2774, Apr. 2021, doi:
10.1016/j.comcom.2022.04.019.
10.3390/s21082774.
[3] F. Amin, A. Ahmad, and G. S. Choi, “Towards trust and friendliness [23] L. M. Verbrugge, “The structure of adult friendship choices,” Social
approaches in the social Internet of Things,” Appl. Sci., vol. 9, no. 1, Forces, vol. 56, no. 2, pp. 576–597, 1977, doi: 10.2307/2577741.
p. 166, Jan. 2019, doi: 10.3390/app9010166.
[24] C. Marche, L. Atzori, and M. Nitti, “A dataset for performance analysis
[4] M. S. Roopa, S. Pattar, R. Buyya, K. R. Venugopal, S. S. Iyengar, and of the social Internet of Things,” in Proc. IEEE 29th Annu. Int. Symp.
L. M. Patnaik, “Social Internet of Things (SIoT): Foundations, thrust Pers., Indoor Mobile Radio Commun. (PIMRC), Sep. 2018, pp. 1–5, doi:
areas, systematic review and future directions,” Comput. Commun., 10.1109/PIMRC.2018.8580830.
vol. 139, pp. 32–57, May 2019, doi: 10.1016/j.comcom.2019.03.009. [25] L. Pu, X. Chen, J. Xu, and X. Fu, “D2D fogging: An
[5] S. Iftikhar et al., “HunterPlus: AI based energy-efficient task schedul- energy-efficient and incentive-aware task offloading framework via
ing for cloud–fog computing environments,” Internet Things, vol. 21, network-assisted D2D collaboration,” IEEE J. Sel. Areas Commun.,
Apr. 2023, Art. no. 100667, doi: 10.1016/j.iot.2022.100667. vol. 34, no. 12, pp. 3887–3901, Dec. 2016, doi: 10.1109/JSAC.
[6] I. U. Din, K. A. Awan, A. Almogren, and B.-S. Kim, “ShareTrust: Cen- 2016.2624118.
tralized trust management mechanism for trustworthy resource sharing in [26] X. Chen, L. Jiao, W. Li, and X. Fu, “Efficient multi-user com-
industrial Internet of Things,” Comput. Electr. Eng., vol. 100, May 2022, putation offloading for mobile-edge cloud computing,” IEEE/ACM
Art. no. 108013, doi: 10.1016/j.compeleceng.2022.108013. Trans. Netw., vol. 24, no. 5, pp. 2795–2808, Oct. 2016, doi: 10.1109/
[7] K. C. Chung and S. W.-J. Liang, “An empirical study of social network TNET.2015.2487344.
activities via social Internet of Things (SIoT),” IEEE Access, vol. 8, [27] X. Chen, “Decentralized computation offloading game for
pp. 48652–48659, 2020, doi: 10.1109/ACCESS.2020.2978151. mobile cloud computing,” IEEE Trans. Parallel Distrib. Syst.,
[8] I.-R. Chen, F. Bao, and J. Guo, “Trust-based service management vol. 26, no. 4, pp. 974–983, Apr. 2015, doi: 10.1109/TPDS.2014.
for social Internet of Things systems,” IEEE Trans. Depend. Secure 2316834.
Comput., vol. 13, no. 6, pp. 684–696, Nov./Dec. 2016, doi: 10.1109/ [28] M. Nitti, L. Atzori, and I. P. Cvijikj, “Friendship selection in the
TDSC.2015.2420552. social Internet of Things: Challenges and possible strategies,” IEEE
[9] L. Atzori, A. Iera, and G. Morabito, “From ‘smart objects’ to ‘social Internet Things J., vol. 2, no. 3, pp. 240–247, Jun. 2015, doi: 10.1109/
objects’: The next evolutionary step of the Internet of Things,” IEEE JIOT.2014.2384734.
Commun. Mag., vol. 52, no. 1, pp. 97–105, Jan. 2014, doi: 10.1109/ [29] F. Amin, R. Abbasi, A. Rehman, and G. S. Choi, “An advanced
MCOM.2014.6710070. algorithm for higher network navigation in social Internet of Things
[10] B. Bhola et al., “Quality-enabled decentralized dynamic IoT platform using small-world networks,” Sensors, vol. 19, no. 9, p. 2007, Apr. 2019,
with scalable resources integration,” IET Commun., to be published, doi: doi: 10.3390/s19092007.
10.1049/cmu2.12514. [30] A. Lakhan et al., “Delay optimal schemes for Internet of Things
[11] G. Jiang, B. Hu, and Y. Wang, “Agent-based simulation of competitive applications in heterogeneous edge cloud computing networks,”
and collaborative mechanisms for mobile service chains,” Inf. Sci., Sensors, vol. 22, no. 16, p. 5937, Aug. 2022, doi: 10.3390/
vol. 180, no. 2, pp. 225–240, Jan. 2010, doi: 10.1016/j.ins.2009.09.014. s22165937.

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

10 IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS

Alireza Souri (Senior Member, IEEE) is currently Asghar Mohammadian is currently a Faculty Mem-
an Associate Professor with the Department of ber with Islamic Azad University, Ilkhchi, Iran. His
Software Engineering, Haliç University, Istanbul, research interests are cloud and edge computing, the
Turkey. Also, he collaborates as a Research Fel- IoT, deep learning, and autonomous vehicle systems.
low at the School of Electrical and Electronics
Engineering, Shandong University of Technology,
Zibo, China. In the scientific community, he has
been listed among the Top 2% Scientist in World
released by Stanford University, USA. In addition,
he has coauthored more than 85 scientific articles
and conference papers in high-ranked journals. His
research interests include formal verification, cloud computing, the Internet
of Things (IoT), and data mining.

Jin Shen received the Ph.D. degree in optical engi-

Yanlei Zhao received the Ph.D. degree in electrical neering from the University of Shanghai for Science
engineering from the Institute of Electrical Engineer- and Technology, Shanghai, China, in 2003.
ing, Chinese Academy of Sciences, Beijing, China, He is currently a Professor with the School of
in 2006. Electrical and Electronic Engineering, Shandong
He is currently working as a Professor with the University of Technology, Zibo, China. His main
School of Electrical and Electronic Engineering, research interests include signal detection and pro-
Shandong University of Technology, Zibo, China. cess, the IoT, and deep learning.
His research interests are the Internet of Things (IoT)
and smart distribution grid.

Mingliang Gao received the Ph.D. degree in com- Eyhab Al-Masri (Senior Member, IEEE) is cur-
munication and information systems from Sichuan rently an Assistant Professor with the Depart-
University, Chengdu, China, in 2013. ment of Computer Science and Systems, School
He is currently working as an Associate Professor of Engineering and Technology (SET), University
with the School of Electrical and Electronic Engi- of Washington Tacoma, Tacoma, WA, USA. His
neering, Shandong University of Technology, Zibo, research interests focus on computational resource
China. His main research interests include computer sharing, resource allocation and optimization, ser-
vision and deep learning. vice migration, the Internet of Things (IoT) security,
and edge AI for the IoT.

Authorized licensed use limited to: Istinye Universitesi. Downloaded on April 21,2023 at 01:19:21 UTC from IEEE Xplore. Restrictions apply.