A fuzzy logic-based model for filtering dishonest recommendations in the Social Internet of Things

Abstract

In the recent year, Internet of Things (IoT) has been adopted in several real-world applications such as smart transportation, smart city, retail, agriculture, smart factory, etc. to make human life more reliable. The integration of social networking concepts into the IoT led to the rise of a new paradigm: the Social Internet of Things (SIoT). In the SIoT environment, the objects are capable of establishing in an autonomous way many social relationships anywhere and anytime with other trusted objects. However, in such environment, objects may provide dishonest recommendations due to malicious reasons such as bad mouthing, ballot stuffing, random opinion, etc. In order to cater these challenges, we propose a new fuzzy logic-based model to filter dishonest recommendations and estimate their trust level based on (1) their values and sending time and the place coordinates and (2) the social relationship parameters of the recommenders. Results prove that our proposed approach is able to detect 100% of the fake Sybil attack and achieves 100% of Recognition Proportion, Sensitivity, Specificity, Accuracy and F1 score and gets 0% of False Negative and False Positive Proportions in presence of up to 90% dishonest recommendations.