Infrastructure-Based Communication Trust Model for Intelligent

Transportation Systems

Malek Lachheb

1 a

, Rihab Abidi

1,2 b

, Nadia Ben Azzouna

1 c

and Nabil Sahli

3 d

Universit

e de Tunis, Institut Sup

erieur de Gestion de Tunis, SMART Lab, Av. de la Liberte, Tunis, Tunisia

Normandy University, UNIROUEN, ESIGELEC, IRSEEM, Av. Galilee, Normandie, France

Computer Science Department, German University of Technology in Oman (GUtech), Muscat, Oman

Keywords:

Trust Model, Communication Trust, Infrastructure-Based Model, Smart Road Signs, Fuzzy Logic, Dempster

Shafer Theory.

Abstract:

Intelligent Transportation Systems (ITS) aim to enhance trafﬁc management through Vehicle-to-Vehicle

(V2V), Vehicle-to-Infrastructure (V2I) and Infrastructure-to-Infrastructure (I2I) communications. However,

the wireless medium and dynamic nature of these networks expose them to security threats from faulty nodes

or malicious attacks. While cryptography-based mechanisms provide security against outsider attacks, the

network remains vulnerable to attacks from legitimate but malicious nodes. Trust models have hence been

proposed to evaluate node and data credibility to make informed security decisions. Existing models are either

vehicle-centric with limited stability due to mobility or infrastructure-based with risks of single points of fail-

ure. This paper proposes a self-organizing, infrastructure-based trust model for securing ITS communication

leveraging Smart Roadside Signs (SRSs). The model introduces a trust-based clustering algorithm using a

fuzzy-based Dempster Shafer Theory (DST). This eliminates dependence on external trusted authorities while

enhancing stability through infrastructure oversight. The decentralized trust formation and adaptive clustering

balance security assurance with scalability. The results of the simulations show that our model is resilient

against on-off attack, packet drop attack, jamming attack, bad-mouthing attack and collusion attack.

1 INTRODUCTION

Intelligent Transportation Systems (ITS) are envi-

sioned to enhance trafﬁc management and coor-

dination through the integration of communication

technologies into transportation infrastructure. ITS

ecosystems comprise interacting vehicles, and road-

side infrastructures. Exchanging information through

interconnections such as Vehicle-to-Vehicle, Vehicle-

to-Infrastructure, and Infrastructure-to-Infrastructure

networks. However, the wireless medium and highly

dynamic nature of vehicular networks expose them to

security threats arising from both malicious attacks

as well as inadvertent faults due to physical failure.

While cryptography-based security mechanisms can

protect against outsider attackers, threats remain from

authorized nodes conducting insider attacks and from

https://orcid.org/0009-0002-5540-7222

https://orcid.org/0000-0002-6108-7854

https://orcid.org/0000-0002-6953-2086

https://orcid.org/0000-0002-9805-6859

faulty nodes transmitting erroneous data.

To address these concerns, trust management

models have been proposed for securing ITS networks

and to aid real-time decision making on the ﬁdelity of

received data and reliability of communication nodes

themselves. The majority of existing trust models are

vehicle-centric, with nodes evaluating others. How-

ever, high mobility of vehicles limited stability in as-

sessment. On the other hand, infrastructure-based

trust models rely on ﬁxed nodes such as Road Side

Units (RSUs) for trust evaluation and Trusted Athori-

ties for credential and certiﬁcates management but re-

main prone to single point failures. Centralized archi-

tectures also pose scalability challenges.

This paper develops a self-organizing

infrastructure-supported trust mechanism for re-

liable ITS networking, called Infrastructure-based

Communication Trust model for Intelligent Trans-

portation Systems (ICT4ITS). The model employs

Smart Roadside Signs (SRS) as distributed infras-

tructure entities to facilitate trust formation through

collaborative assessment (Abidi et al., 2023). The

Lachheb, M., Abidi, R., Ben Azzouna, N. and Sahli, N.

Infrastructure-Based Communication Trust Model for Intelligent Transportation Systems.

DOI: 10.5220/0012738700003702

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 10th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2024), pages 513-521

ISBN: 978-989-758-703-0; ISSN: 2184-495X

513

hybrid architecture balances scalability, with stability

and ensures the resiliency for security requirements.

In what follows, we deﬁne our main contributions:

(1) we introduce an infrastructure-based trust model

to ensure the stability of the network; (2) propose a

self-organizing Manager SRS election and monitor-

ing to enhance the scalability of the network, (3) and

employ a fuzzy-based Dempster Shafer Theory tech-

nique to evaluate the trustworthiness of the SRSs. The

rest of the paper is organized as follows: section 2

breiﬂy reviews the existing trust models. Section 3

introduces the architecture of the model, highlights

its overall workﬂow and deﬁnes its considered pa-

rameters. The detailed description of the model is

presented in section 4. The results are discussed in

section 5. Section 6 concludes the paper.

2 LITERATURE REVIEW

Securing communications and establishing trust be-

tween nodes is essential for the adoption of ITS ap-

plications. In what follows, we review the state of the

art in order to analyse recent research on trust models

and security mechanisms for enabling reliable com-

munication in ITS.

Eunice and Juvanna proposed a secured multi-hop

clustering protocol that uses a weighted voting-based

cluster head election (Eunice and Juvanna, 2019).

Nodes monitor behavior locally to evaluate trust lev-

els. While providing authentication and integrity, the

decentralized approach can introduce communication

overhead. Kerrache introduced a trust-aware archi-

tecture using named data networking to form trusted

vehicle groups (Kerrache, 2022). A trusted third

party evaluates nodes based on credentials, recom-

mendations and plausibility checks. However, fre-

quent key distribution may lead to scalability issues.

Gupta et al. developed an enhanced beacon trust

management model integrated with a clustering pro-

tocol (Gupta et al., 2023). Malicious nodes are de-

tected using plausibility checks on periodic beacons.

The authors use vector points that represent the ve-

hicle’s position, speed, and driver direction sent in

the beacon message. On the one hand, a centralized

server is employed to analyse and compare the mes-

sages and detect malicious nodes using predeﬁned

thresholds sets. On the other hand, vehicles’ densities

and velocities change rapidly due to the dynamic na-

ture of VANET. Thus, the centralized threshold-based

mechanism lacks adaptability to dynamic topologies.

Kchaou et al. presented a secured clustering tech-

nique using proxy re-encryption to enable authorized

data sharing between cluster members. But the single

cluster head is a single point of failure (Kchaou et al.,

2018). Hasrouny et al. used an opinion dynamics-

based model for establishing trust relations between

nodes to select reliable group leaders (Hasrouny et al.,

2019). Alsuhli et al. introduced a double-headed clus-

tering structure to improve resilience to targeted at-

tacks on cluster heads (Alsuhli et al., 2019). The de-

centralized approach enhances reliability but latency

and overhead needs evaluation. Fatemidokht and Raf-

sanjani proposed a quality of service and monitoring-

based clustering algorithm that detects misbehaving

nodes (Fatemidokht and Kuchaki Rafsanjani, 2020).

The multi-metric decision enables adaptive clustering

based on dynamic contexts.

Scalability constitutes a major constraint for the

communication trust models. For instance, the au-

thentication schemes proposed in (Eunice and Ju-

vanna, 2019) and (Kerrache, 2022) arise the scalabil-

ity issue, due to the increase of communication over-

head and latency. Moreover, the stability concern oc-

curs, with the lack of awareness of the dynamic na-

ture of the ITS environment, the high speed move-

ment of the vehicles, and by employing centralized

architectures that increase the risk of single point fail-

ure, such as in (Gupta et al., 2023), and (Kchaou et al.,

2018). On the one hand, most recent works deployed

decentralized approaches for self-organized trust es-

tablishment between vehicles without dependence on

external infrastructure. While this method provides

better scalability and avoids single points of failure, it

lacks the stability of the network. On the other hand,

most of the infrastructure based models rely on a sin-

gle trust authority and they are mainly used for certiﬁ-

cate and credential managements, which ensures sta-

bility but lacks scalability and increases the risk of a

single point failure. Exploring models that introduce

a trade-off between the scalability and stability of the

network, while ensuring the robustness of the com-

munication, maybe a promising research direction to

meet the requirements of ITS applications.

In our proposed model, we combine the infras-

tructure based model with the self organizing trust,

by introducing the Smart Road Signs, where the trust

establishment is realized based on cooperation of the

SRSs. This combination ensures the scalability and

stability of the network while securing the network

communication and increasing the robustness of the

model.

3 SYSTEM DESIGN

This section introduces the architecture of the pro-

posed model, the main phases of the trust workﬂow,

VEHITS 2024 - 10th International Conference on Vehicle Technology and Intelligent Transport Systems

514

and the considered attacks and parameters.

3.1 ICT4ITS: Architecture

In this paper, we introduce a decentralized communi-

cation trust model for ITS leveraging different com-

ponents including Smart Road Signs (SRSs) and a

watchdog module. Figure 1 outlines the overall sys-

tem architecture.

Figure 1: General architecture of the ICT model.

We propose partitioning the trafﬁc road system

into regional areas constituting sets of connected

roads with embedded regulatory nodes. These over-

sight nodes, consisting of SRSs, govern trafﬁc data

ﬂows within their regional roads. Speciﬁcally, SRSs

are signs equipped with digital displays that provide

real-time trafﬁc notiﬁcations to drivers, enabled by

processing capabilities and internet connectivity. We

conﬁgure SRSs in a hierarchical structure with two

types: Manager SRSs and Subordinate SRSs, as de-

ﬁned in (Sahli et al., 2022). Managers oversee Subor-

dinates within their supervised region. Furthermore,

our framework incorporates watchdog smodules as

described by Siddiqua et al. (Siddiqua and Jahan,

2022) and Akwirry et al. (Akwirry et al., 2022). The

watchdog inclusion constitutes an important element

that boosts reliability within our model. These over-

seer components are deployed within a Trusted Ex-

ecution Environment (TEE) to securely monitor in-

coming and outgoing trafﬁc of other entities inside

the transmission range. In particular, the watchdog

can overhear network activity among nodes as long

as they reside within proximity and save them in its

pool. By eavesdropping on packet routing of nearby

SRSs, the watchdog veriﬁes whether packets are cor-

rectly forwarded to the subsequent node. Moreover,

according to Akwirry et al. (Akwirry et al., 2022),

watchdogs may be leveraged to evaluate transmission

trustworthiness levels and perform computational an-

alyzes. Thus, the watchdog’s omnidirectional moni-

toring capability facilitates the calculation of accurate

trust values of SRSs within its region.

We propose a two-layer trust evaluation. In the

lower layer trust evaluation, Subordinate SRSs use

fuzzy logic to estimate the trustworthiness of neigh-

bouring SRSs. The output of the lower layer evalu-

ation of each Subordinate SRS is transmitted to the

Manager SRS for aggregation. Accordingly, Man-

ager SRS applies the DST for trust aggregation in the

higher level trust evaluation.

3.2 ICT4ITS: System Overview

The SRSs are exposed to potential security threats and

they may even transmit erroneous information. Since

Manager SRSs aggregate the evaluations of the Sub-

ordinate SRSs and dictate regional operations, their

security impacts the overall system functionality. Ac-

cordingly, ensuring trustworthy Manager SRSs con-

stitutes critical network vulnerability. We propose a

trust framework centred on SRS behavioural analy-

sis to enhance communications security within SRS

nodes. The proposed model enables classiﬁcation of

SRS nodes as either malicious or honest based on

observed metrics. The proposed model aims to es-

tablish a stable and trusted SRS cluster through two

key phases. Firstly, during the election phase, SRSs

participate in a competitive election process to se-

lect a Manager SRS. Subsequently, in the Manager

SRS Observation phase, Subordinate SRSs exchange

packets with the Manager to report trafﬁc state infor-

mation. Theses packets are then used to continually

evaluate cluster components behaviour based on the

quality of the communication.

The election relies on an initial trust assessment

across SRSs. The SRSs exchange a designated num-

ber of packets. The process runs in a bi-directional

way to enable mutual evaluations between all SRSs.

Then, each SRSs feed Quality of Service indicators

(QoS), stored in the pool of the watchdog module, as

inputs to its fuzzy inference system to estimate the

trustworthiness of neighbouring SRS nodes.

Upon completing the peer trust evaluations via

fuzzy logic, the integrated watchdog aggregates the

trust level outputs to identify the most trusted SRS to

appoint as Manager.

For Manager SRS observation, our hierarchical

topology minimizes packet exchange. Subordinate

SRSs forward packets to the Manager, similarly to

the election process. Concurrently, the watchdog

records communications while Subordinates evaluate

each other through fuzzy logic, utilizing observed pa-

rameters as inference inputs. Subordinates broadcast

assessments to the Manager, which aggregates them

using Dempster Shafer Theory (DST). The Manager

relays the integrated evaluation back to Subordinates.

Subordinate SRSs compare the Manager SRS’s

trust evaluations against locally computed assess-

ments. If the Manager’s assessed trust value for a

Infrastructure-Based Communication Trust Model for Intelligent Transportation Systems

515

monitored node diverges from the Subordinate’s eval-

uation by an established acceptable threshold range,

the Subordinate deems the Manager’s judgment as

honest. However, if the Manager’s rating exceeds the

permissible differential gap (maximum trust value de-

viation tolerated), the Subordinate may determine the

Manager has acted maliciously or erroneously in an

inaccurate manner. Through this comparative anal-

ysis, Subordinates can check and balance the Man-

ager’s evaluations based on ground truth local obser-

vations. Thereby, the integrity of centralized hierar-

chical decisions gets enhanceed by distributed over-

sight of lower-level SRSs. Accordingly, if more than

2/3 of the Subordinates estimate that the Manager has

acted maliciously, a new election phase is triggered.

3.3 ICT4ITS: Attack Model

In this paper, we consider the following communica-

tion related attacks (Shetty and Manjaiah, 2022).

• Packet Drop Attack (PDA). Malicious node

drops a number of packets, intentionally or un-

intentionally, disrupting communication and lead-

ing to packet loss.

• Jamming Attack (JA). Malicious nodes inject

multiple packets into the network to overcharge

it, causing network congestion.

• Bad-Mouthing Attack (BMA). Malicious nodes

collude to ruin the reputation of well-behaved

nodes by providing fake negative feedback, un-

dermining trust and credibility.

• On-Off Attack (OOA). Malicious nodes provide

random attacks by injecting erroneous data arbi-

trarily, disrupting communication intermittently.

• Collision Attack (CA). Multiple nodes collabo-

rate to harm the network by intentionally perform-

ing attacks, interfering with legitimate message

transmission and causing data corruption or loss.

3.4 ICT4ITS: Considered Parameters

In the proposed model we consider the factors that

affect the communication performance between the

SRSs. Accordingly, we use packet delivery rate,

throughput, end-to-end delay, and error rate as param-

eters to evaluate the trustworthiness of the SRSs. In

what follows, we deﬁne theses parameters:

• Packet Delivery Rate (PDR): Deﬁned as the ra-

tio between effectively received packets versus to-

tal packets transmitted. Higher PDR correlates to

more reliable delivery, as larger discrepancies in-

dicate potential disruptions degrading availability:

PDR =

Total Received Packets

Total Sent Packets

(1)

• Throughput: Throughput quantiﬁes the amount

of data successfully conveyed across the network

over a set time window. Higher throughput en-

ables dependent real-time services by promoting

responsiveness:

Throughput =

Total Transmitted Packets

Total Time

(2)

• End-to-end delay: Represents packet travel time

from source to recipient. Low latency boosts suit-

ability for time-critical transportation applications

sensitive to lag:

End-to-end delay =

∑

(Receive Time−Send Time)

Total Received Packets

(3)

Error Rate: It measures the proportion of packets

lost compared to the total packets transmitted over

a speciﬁed time period. Lower error rates signal

heightened delivery dependability:

Error Rate =

Number o f Lost Data Packets

Total Sent Packets

(4)

• Number of packets: It refers to the number of

times a task enters a send sleep state awaiting the

transmission of packets to the destination node.

In this parameter, we consider that the nodes ex-

hibit a restricted capacity permitting only singular

packet broadcasts at discrete times, necessitating

interim sleep states between each networked send

operation.

The deployed watchdog module is dedicated to

passively monitor communications and obtain ground

truth recordings of the before-mentioned parameters

that are used for trust computation. In fact, the PDR

and throughput metrics serve as indicators to detect

PDA. The Number of Packets metric aids in identi-

fying OOA. Furthermore, end-to-end delay and error

rate metrics are instrumental in detecting JA.

4 ICT4ITS: OPERATIONAL

PHASE DESCRIPTION

As vehicle networks exhibit inherently dynamic

and uncertain environments, fuzzy logic pro-

vides VANET-based systems with ﬂexible, adaptive

decision-making capabilities by leveraging human ex-

pert knowledge. This suitability for handling inex-

act real-world inputs endows fuzzy architectures with

VEHITS 2024 - 10th International Conference on Vehicle Technology and Intelligent Transport Systems

516

an apt trust scoring mechanism for SRS components

reliant on variable network transmission reliability.

Fuzzy logic is composed mainly of three steps deﬁned

as follows (Zadeh, 2004): (1) fuzziﬁcation, where

crisp input values are translated into degrees of mem-

bership across fuzzy sets using functions like sig-

moid, trapezoidal, Gaussian, or triangular (Pedrycz,

1994); (2) inference system, which involves mapping

fuzziﬁed inputs to outputs through if-then rule sets;

(3) defuzziﬁcation, where fuzzy output sets are quan-

tiﬁed into singular or multiple values using methods

like centroid, bisector, or mean of maxima (Saade and

Diab, 2000).

In our model, we adopt the triangular presenta-

tion as fuzziﬁer due to its simplicity and efﬁciency

(Souissi et al., 2023). We use the inference system

proposed in (Umoren et al., 2019). Finally, we apply

centroid technique based on its demonstrated effec-

tiveness for consolidating expressions of incomplete

knowledge into decisive score reporting (Saade and

Diab, 2000). The Subordinate SRSs’ fuzzy systems

take as an input the packet delivery rate, throughput,

number of packets, end-to-end delay, and error rate.

In what follows, we present the linguistic variables

associated to each input: Packet delivery rate [low,

moderate, high], throughput [very low, low, moder-

ate, high], number of packets [less, average, more],

end-to-end delay [short, normal, long], and error rate

[very low, low]. The linguistic terms associated to

the output are: trustworthiness [very low, low, moder-

ate, high]. Figures from 2-6 represent the membership

functions of the input variables and ﬁgure 7 presents

the membership function of the output variable.

Figure 2: Fuzzy classes and membership functions for the

packet delivery rate.

Figure 3: Fuzzy classes and membership functions for the

number of packets.

Our model adopts Mamdami fuzzy inference for

mapping the input parameters to the trustworthiness

of the SRSs using a fuzzy rule-based system (Rahim

Figure 4: Fuzzy classes and membership functions for the

throughput.

Figure 5: Fuzzy classes and membership functions for the

end-to-end delay.

Figure 6: Fuzzy classes and membership functions for the

error rate.

Figure 7: Fuzzy classes and membership functions for the

output.

et al., 2017). Our engine utilizes the Mamdani min-

max implication approach for discerning rule conse-

quences. We used the rule base proposed in (Umoren

et al., 2019). However, this rule base relies primar-

ily on the detection of packet loss and lacks spe-

ciﬁc representations that characterize jamming at-

tacks. Accordingly, we added the missing rules de-

veloped through analysis of analytical data collected

from simulated jamming attack scenarios. Table 1

describes the structure of the IF-THEN of the addi-

tional clauses. Thereafter, defuzziﬁcation phase is ex-

ecuted. We adopt the widespread Centroid method

which ﬁnds the center of gravity (COG) slice-point

equally partitioning the aggregation mass for maxi-

mum representativeness, as follows:

CoG =

x · µ

(x)dx

(5)

Where µ

is the set of membership functions of the

Infrastructure-Based Communication Trust Model for Intelligent Transportation Systems

517

fuzzy set A and x is the variable over the universe of

discourse X.

We employ Dempster-Shafer theory (DST) as

an aggregator for consolidating potentially disparate

trust evaluations estimated by Subordinate SRSs

nodes into a uniﬁed assessment. In fact, we lever-

age DST for modelling uncertain reasoning. As an

evidence-based mathematical approach, DST proves

useful in decision-making scenarios exhibiting in-

complete or conﬂicting insights. To this end, we con-

vert the fuzzy outputs of the lower layer trust evalua-

tion to belief functions.

Let A be the output set of a subordinate SRS’s

fuzzy inference system, where A

, A

, and A

are

the elements of the fuzzy output set obtained from

a fuzzy inference system, that represents the trust-

worthiness estimated by one subordinate SRS of one

of the neighbouring SRSs. We convert each fuzzy

output set A

into a belief function m

using a map-

ping function f

that assigns Basic Probability Assign-

ments (BPAs) based on the degree of membership of

each element in the output set:

(x) = f

(µ

(x)), (6)

where x is an element in the frame of discernment.

In this context, the frame of discernment is the set A

that encompasses all possible linguistic terms that the

output variable, the trustworthiness, can take. A

the linguistic term of the output of the fuzzy system,

with terms [very low, low, moderate, high]. µ

(x)

represents the membership function of each term.

Therafter, we combine the belief functions

, m

. . . m

, representing the BPAs of

each belief function m

of the SRSs, using DST com-

bination’s rule as follows:

CT = m

⊕ m

⊕ . . . ⊕ m

(7)

Where CT is the curent trust value and n is the num-

ber of the SRSs. The fuzzy-based DST provides

a pathway for consolidating distributed evaluations

under imprecise conditions. Overall, this fuzzy-to-

evidential pipeline supports coherent centralized trust

evaluation within decentralized ITS environments by

combining localized uncertainty management with

global consistencies. The aim of this combination is

to increase the accuracy of the evaluation and enhance

the robustness of the network. After evaluating the

current trustworthiness of each SRS using the fuzzy-

based DST, we update their trustworthiness using a

dynamic weighted sum as shown in Equation 8:

U pdated trust value = α · CT + β· HT (8)

Where CT is the newly assessed trust value and HT is

the historical trust value.

This consolidates the recent evaluation with his-

toric trustworthiness to balance temporary ﬂuctua-

tions with overall trends. The α and β parameters de-

termine weighting concreted to new versus old evalu-

ations respectively. We implement an adaptive tun-

ing approach for α and β assignments to provide

custom credibility response rates per SRS based on

past performance proﬁles. Well-behaved nodes are

assigned higher α prioritizing current evaluations to

quickly improve standing. However, for SRSs ex-

hibiting bursts of maliciousness, higher β maintains

inﬂuence of prior windows to dampen volatility ef-

fect. Accordingly, if the old trust value of the SRS is

higher than 0.6 we assign 0.7 to α and 0.3 to β. If the

old trust value ranges between [0.4, 0.6], we assign

a neutral weight to α and β equals to 0.5. Finally,

if the old trust value is lower than 0.4, we prioritize

higher β value equals 0.7 and lower α equals to 0.3.

Thereby, misbehaving SRSs require more consistent

integrity demonstrations before trust value upgrades.

This strategy strengthens resilience against strategic

oscillation tactics aimed at brieﬂy feigning good be-

haviour to swiftly regain network standing after at-

tacks. Through tailored weighting, we promote fair

credibility aggregation while preventing exploitation.

5 PERFORMANCE EVALUATION

In this section, we present the simulation setup and

and discuss the results. In order to generate the

data, we used Network Simulator (NS2). NS2 is an

open-source, event-driven simulator, designed specif-

ically for research in computer communication net-

works. Moreover, we used Generator Network Sim-

ulator (GNS2) for the script generation, which is a

software based on the drag and drop technique to gen-

erate the Tool Command Language (TCL) for NS2 of

the deﬁned scenario.

Our simulation scenario comprises 10 nodes, rep-

resenting our SRSs, situated within shared wireless

transmission range modeling a decentralized network

within the same cluster. The Nodes perform as the

data sink and/or active sources. Trafﬁc generation

occurs via File Transfer Protocol (FTP) encapsulated

through Transmission Control Protocol packet (TCP)

for reliability. We conﬁgure TCP agents on each node

to handle connection setup, data transfer, ﬂow con-

trol and event handling. FTP is a protocol used for

transferring ﬁles between computers on a network. It

is not characterized by a constant bit rate, but rather

involves the transfer of ﬁles, which may vary in size

and number, and may take varying amounts of time

to transmit. During simulation execution, we gather

VEHITS 2024 - 10th International Conference on Vehicle Technology and Intelligent Transport Systems

518

Table 1: Rules base structure.

Packet delivery Throughput Number of packets Delay Error rate Trust

high high Average long low low

high high more long low low

high high Average normal low low

high high more normal low low

high very low Average long low low

high very low more long low low

high very low Average normal low low

high very low more normal low very low

high high Average long low low

high high more long low very low

high high Average normal low low

high high more normal low low

high very low Average long low low

high very low more long low very low

high very low Average short very low high

high very low Average normal low medium

high very low more normal low very low

high high Average long low low

moderate high more long low very low

moderate high Average normal low low

moderate high more normal low very low

moderate very low Average long low low

moderate very low more long low very low

moderate very low Average normal low low

moderate very low more normal low low

moderate high Average long low low

moderate high more long low low

moderate high more normal low low

moderate high Average normal low low

per-ﬂow metrics including delay, error rate, delivery

rate, throughput and absolute packet deliveries.

In what follows, we discuss the numerical results

of the trust model across different scenarios to assess

its robustness in the presence of different attacks and

the stability of the cluster formation. In all subsequent

ﬁgures, the x-axis denotes the number of iterations,

while the y-axis represents the trust value of the SRSs.

5.1 Model’s Robustness Evaluation

In order to evaluate the robustness of our model, we

examine its performance in the presence of different

percentages of malicious nodes and various attacks.

In the ﬁrst scenario, we test the resiliency of our

model against on-off attack and packet drop attack.

Accordingly, we track the evolution of the trust value

of a well-behaving node and a malicious node per-

forming packet drop attack within on-off mode. The

results of the experimentation are presented in ﬁgure

8. The malicious SRS performed the PDR in the fol-

lowing iterations: 2, 4-9, 13-15, 18-20, 24, 27-29.

In the second scenario, we test the resiliency of

the proposed model against on-off attack and bad

mouthing attack, as presented in ﬁgure 9. In this sce-

nario, the malicious SRS performed attacks at itera-

tions number: 1-5, 11-5, 22-25.

In both scenarios, the malicious SRS and the be-

nign SRS have initial trust value equals to 0.5. We

notice in ﬁgure 8 and 9, that the trust value of the

malicious node decreases rapidly whenever it starts

performing the malicious attacks. However, when it

readopts a good behaviour, the trust value is slowly

growing compared to the drop rate when it performs

attacks. This is due to the adaptive weight of the trust

update. In fact, when the malicious nodes presents a

good behaviour after being malicious the model uses

higher β value. Accordingly, the trust updates rely

more on old trust evaluation, which requires the ma-

licious node to present a good performance for a long

time window to upgrade its trust value. This mecha-

nism allows our model to track alternative behaviour

of malicious nodes and detect their attacks.

In the third scenario, we test the robustness of

our model against packet drop attack, jamming attack,

and bad mouthing attack, as shown in 10, 11, and 12,

respectively. Unlike the ﬁrst and second scenario, in

this scenario the malicious SRS are constantly per-

forming attacks. As shown in the ﬁgures, the benign

SRSs quickly converge to a high trust value. Simi-

Infrastructure-Based Communication Trust Model for Intelligent Transportation Systems

519

Figure 8: Trustworthiness of a well-behaving SRS and a

malicious SRS performing OOA and PDA.

Figure 9: Trustworthiness of a well-behaving SRS and a

malicious SRS performing OOA and BMA.

larly, the malicious SRSs converge rapidly to a low

trust value after few iterations of performing attacks.

This can be explained by the employment of a fuzzy-

based DST for trust computation. This mechanism

allows the model to increase the accuracy of the eval-

uation by handling the uncertainty of the data. The

two-layer evaluation enhance the visibility of the en-

vironment by combining more evidence to ensure the

accuracy of the evaluation and the robustness of the

network. Moreover, the adaptive trust update mech-

anism ensures that the update of the trust values of

benign SRS consider recent trust evaluation for rapid

growth. On the other hand, it ensures that the trust up-

date for the malicious SRSs considers more old trust

values, guaranteeing a quick decrease of their trust

value among the iterations.

Figure 10: Trustworthiness of a well-behaving SRS and a

malicious SRS performing PDR.

Figure 11: Trustworthiness of a well-behaving SRS and a

malicious SRS performing jamming attack.

Figure 12: Trustworthiness of a well-behaving SRS and a

malicious SRS performing BMA.

5.2 Cluster Stability Evaluation

In the last experimentation, we examine the stabil-

ity of the cluster by examining the time laps before

starting a new Manager election phase. Moreover,

we examine the trustworthiness of the Manager in the

meantime. In this scenario, we suppose that the Man-

ager is well-behaving and in the ﬁrst 15 iterations, we

inject 1/3 of the number of the Subordinate SRSs per-

forming BMA. After iteration 15, we increase the per-

centage of the malicious nodes to 2/3 out of the total

subordinate SRSs. The results of the experimentation

are plotted in ﬁgures 13, and 14, respectively. The

ﬁgures show that whenever the number of the ma-

licious nodes performing BMA is less than the 2/3

of the Subordinate SRSs, the fake feedbacks do not

affect the stability of the cluster and the trustworthi-

ness of the Manager. However, after iteration 15 the

trustworthiness of the Manager decreases, although

its is a benign SRS and a new election process is trig-

gered. This can be explained by the fact that when

the number of the malicious SRSs performing BMA

increases, the combination of their evaluations affects

the ﬁnal evaluation. However, it is worth mentioning

that this case scenario is unrealistic and it is implausi-

ble to have more than 2/3 of the SRS to become com-

promised simultaneously.

Figure 13: Stability of the Manager SRS.

Figure 14: Trustworthiness of the Manager SRS.

VEHITS 2024 - 10th International Conference on Vehicle Technology and Intelligent Transport Systems

520

6 CONCLUSIONS

The paper introduces a decentralized trust manage-

ment model for electing trustworthy Manager SRS

and forming stable clusters. It ensures continuous

cluster monitoring by tracking Manager SRS behav-

ior via collaborative assessment among subordinate

SRSs. The proposed model integrates the stabil-

ity of infrastructure-based evaluation with the scal-

ability and resilience of decentralized trust architec-

ture. Self-organizing clustering dynamically selects

trusted Managers, reducing communication overhead

and eliminating single point failures. Trust compu-

tation, using fuzzy logic and DST, increases the ac-

curacy of trust evaluation by handling the uncertainty

of the input and the conﬂicted outputs of the SRSs.

Simulation-based evaluation reveals resilience against

OOA, PDA, JA, BMA, and CA but struggles with col-

lusion and BMA when malicious SRSs exceed 2/3 of

the total. Despite potential computational complex-

ity, combining fuzzy inference with DST strength-

ens the network, offering robustness and ﬂexibility

for uncertain and conﬂicting evidence. Future work

includes evaluating response time and incorporating

social metrics like honesty and cooperativeness to en-

hance trust evaluation accuracy.

ACKNOWLEDGEMENTS

The research leading to these results has received

funding from the Ministry of Higher Education, Re-

search and Innovation of the Sultanate of Oman under

the Block Funding Program. Block Funding Agree-

ment No [BFP/RGP/ICT/22/327].

The English quality of the paper is enhanced using

the AI assistant Claude (Anthropic, 2023).

REFERENCES

Abidi, R., Sahli, N., Trojet, W., Azzouna, N. B., and Hob-

los, G. (2023). An infrastructure-based trust manage-

ment framework for cooperative its. In VEHITS, pages

329–336.

Akwirry, B., Bessis, N., Malik, H., and McHale, S. (2022).

A multi-tier trust-based security mechanism for ve-

hicular ad-hoc network communications. Sensors,

22(21):8285.

Alsuhli, G. H., Khattab, A., Fahmy, Y. A., et al. (2019).

Double-head clustering for resilient vanets. Wireless

communications and mobile computing, 2019.

Anthropic (2023). Claude - AI assistant. https://www.an

thropic.com/claude. [Online; accessed 23-February-

2023].

Eunice, K. S. and Juvanna, I. (2019). Secured multi-

hop clustering protocol for location-based routing in

vanets. International Journal of Advanced Computer

Science and Applications, 10(4).

Fatemidokht, H. and Kuchaki Rafsanjani, M. (2020). Qmm-

vanet: An efﬁcient clustering algorithm based on qos

and monitoring of malicious vehicles in vehicular

ad hoc networks. Journal of Systems and Software,

165:110561.

Gupta, C., Singh, L., and Tiwari, R. (2023). Malicious

node detection in vehicular ad-hoc network (vanet)

using enhanced beacon trust management with clus-

tering protocol (ebtm-cp). Wireless Personal Commu-

nications, 130(1):321–346.

Hasrouny, H., Samhat, A. E., Bassil, C., and Laouiti,

A. (2019). Trust model for secure group leader-

based communications in vanet. Wireless Networks,

25:4639–4661.

Kchaou, A., Abassi, R., and El Fatmi, S. G. (2018). To-

wards a secured clustering mechanism for messages

exchange in vanet. In 2018 32nd International Con-

ference on Advanced Information Networking and Ap-

plications Workshops (WAINA), pages 88–93.

Kerrache, C. A. (2022). A trust-aware cluster-based com-

munication architecture for vehicular named data net-

working. ITU J. on Future and Evolving Technologies.

Pedrycz, W. (1994). Why triangular membership functions?

Fuzzy Sets and Systems, 64(1):21–30.

Rahim, R. et al. (2017). Comparative analysis of member-

ship function on mamdani fuzzy inference system for

decision making. In Journal of Physics: Conference

Series, volume 930, page 012029. IOP Publishing.

Saade, J. and Diab, H. (2000). Defuzziﬁcation techniques

for fuzzy controllers. IEEE Transactions on Sys-

tems, Man, and Cybernetics, Part B (Cybernetics),

30(1):223–229.

Sahli, N., Trojet, W., Zhang, Z., and Abdallah, N. O.

(2022). Towards a network of dynamic message signs

for congestion alerting. Computing and Informatics,

41(2):609–626.

Shetty, S. R. and Manjaiah, D. H. (2022). A comprehen-

sive study of security attack on vanet. In Sharma, N.,

Chakrabarti, A., Balas, V. E., and Bruckstein, A. M.,

editors, Data Management, Analytics and Innovation,

pages 407–428, Singapore. Springer Singapore.

Siddiqua, F. and Jahan, M. (2022). A trust-based malicious

rsu detection mechanism in edge-enabled vehicular ad

hoc networks. arXiv preprint arXiv:2208.05680.

Souissi, I., Abidi, R., Azzouna, N. B., Berradia, T., and

Said, L. B. (2023). Ecotrust: A novel model for en-

ergy consumption trust assurance in electric vehicular

networks. Ad Hoc Networks, 149:103246.

Umoren, I. J., Asuquo, D. E., Gilean, O., and Esang, M.

(2019). Performability of retransmission of loss pack-

ets in wireless sensor networks. Comput. Inf. Sci.,

12(2):71–86.

Zadeh, L. A. (2004). A note on web intelligence, world

knowledge and fuzzy logic. Data & Knowledge Engi-

neering, 50(3):291–304. Special jubilee issue: DKE

50.

Infrastructure-Based Communication Trust Model for Intelligent Transportation Systems

521