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

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.