A resilient Trust Management framework towards trust related attacks in the Social Internet of Things

Abstract

As a huge number of heterogeneous smart things in the Social Internet of Things (SIoT) paradigm are connected to exchange services, security principles are needed to ensure safe communication in the network. From this perspective, various Trust Management models have been set forward in the literature in order to assess nodes trustworthiness and therefore to avoid malicious connections in the SIoT network. However, these models still lack practical solutions to address SIoT trust related attacks performed towards the trust evaluation process. In this regard, the present work stands for an in-depth study on the effectiveness of the trust attack management, as it is integrated into the trust model, with the aim of classifying the nodes behaviors more accurately in order to ensure secure connections in the network. For this reason, nodes behaviors analysis, based on Machine Learning (ML) algorithms, is carried out within our trust assessment process. Our objective lies in limiting interactions with not only poor service provider, but also with attacker nodes. The proposed trust method’s acquired outcome reveals that our introduced TM model demonstrates an improved efficiency in identifying malicious nodes compared to a trust classification algorithm without nodes behaviors evaluation. Experimentation conducted within a simulated dataset based on real social data through Cooja; corroborates the efficiency of our proposed solution in terms of coping with simulated trust related attacks performed by malicious nodes.

Introduction

The Social Internet of Things (SIoT) paradigm corresponds to an evolutionary Internet of Things (IoT) where objects are transformed from classical smart devices to social objects, which are able to autonomously interact with their social environment [1] in order to produce wealthier and more varied data for more complex services. In this environment, SIoT objects can be either trustworthy by producing good services, or untrustworthy by producing poor services. This refers basically to the vulnerability of easily accessible objects in unprotected zones to various attacks. In fact, some of nodes try to produce misbehaving behaviors to damage the system performance. Thus, they introduce various types of cybersecurity attacks called : Distributed denial of services (DOS), malicious control, wrong, data type probing, malicious operation, etc. For this reason, humans remain careful and cautious of the disclosure of their data [2], [3]. Thus, to ensure user’s satisfaction and boost SIoT system performance, it is crucial for nodes to evaluate one another before exchanging services. In this case, the issue of Trust Management (TM) solutions for SIoT system becomes intrinsic.

So far, the works carried out in the field of trust have whetted a special interest in the elaboration of TM systems for tackling related security problems [4], [5], [6]. It allows humans to face and overcome their fears as well as uncertainties about the use of SIoT applications. To ensure reliable services, numerous studies handling trust mechanisms have been elaborated in various areas such as: peer-to-peer (P2P) [7], [8], Social Network (SN) [9], [10], [11], Cloud [12], [13], [14] and IoT [15], [16], [17], [18]. However, each of them has its limitations (lack of consideration for nodes’ social features, neglect of service exchange constraints between nodes, etc.), which makes it not efficient enough to be applied into the SIoT systems. Within this framework, several works have been oriented to resolving these gaps. As a matter of fact, they subsequently introduced different TM models that are mainly dedicated to SIoT systems. They considered different features and methods to compute the trust in order to identify which nodes represent the best service providers (SP).

Unlike cybersecurity attacks, attackers nodes resort to propagate other forms of attack called “Trust Related Attacks” TRAs to fake the assessed trust [1]. They aim to share fake recommendations and fake quality service to increase or decrease the trust value for its own benefit so as to enhance their chance to be chosen as good SPs. Basically, most of the proposed TM models are interested in producing a trust score to rank the network node in order to identify trusted ones.

However, in some cases, this score cannot be reliable when it is the result of certain malicious behaviors “TRA”. Therefore, it is crucial to identify untrustworthy nodes owing to the fact that some nodes resort to misbehaving collaborations in order to disrupt the TM system functionality [19].

To this end, some of the existing works in this field [16], [20], [21], [22], [23], proposed different metrics and strategies to handle malicious trust attacks. They focused on identifying malicious attackers nodes in the network to ensure trustworthy communications and qualified services. However, we note that researchers tackling this problem statement did not consider all forms of TRA that can be launched in a SIoT network. This is quite critical as each TRA can be more malicious to another, especially in a collaborative network such as SIoT where trust is based on exchanged recommendations and services.

Based on these limitations, we propose to define a resilient TM model for the SIoT which allows us to detect and identify the various dangerous behaviors resulting in TRA that may occur in the SIoT network, as well as offer relevant trust metrics for detecting them. Unlike other works, we are interested in our model in both attack detection and nodes rating with a trust score to ensure safe communications.

These metrics permit not only to detect and identify TRA, but also to rate SIoT nodes based on a trust score. Unlike other works, we attempt to deal with all the existing TRA that can occur in a SIoT network as each attack can be more threatening than the other one. Thus, our contributions can be summarized as follows:

• The computation of trust SIoT nodes’ score which is based on multiple metrics defined in the trust composition step deriving from social and quality of service nodes’ descriptions.

• The development of the proposed model of Magdich et al. [24] which rests on adding an attack detection module which allows not only the discovery of malicious collaborations, but also the identification of the kind of the performed attack in order to handle them in the future. The identification of malicious TRA is based on node behaviors analysis grounded on Machine Learning algorithm.

• The isolation and the avoidance of misbehaving collaborations which relies on providing users with a SIoT nodes classification into : trustworthy, untrustworthy (while producing unqualified service) and attacker ; to enact a reliable system.

• The simulation of the different SIoT environment, which involves several types of exchange TRA, based on different real social datasets, while overcoming the limits of available SIoT dataset available.

This paper is laid out in the following. In Section 2, we exhibit most of the proposed works focusing on TM to clarify the problem statement. Section 3 displays how trust is computed and managed between peers and clarifies in details the different steps as well as features managed by the attack detection module to analyze malicious behaviors of such SIoT nodes. Section 4 identifies the experiments settings to assess our method referring to state-of-the-art works. Finally, Section 5 wraps up the conclusion.