An intelligent fuzzy logic based-trust system in underwater acoustic sensor networks

Due to the ongoing progress in ocean exploration, underwater acoustic sensor networks (UASNs) have become a significant focus of research. However, the inherent openness and lack of supervision in these networks expose them to various security threats. Thus, efficient and reliable security systems are very necessary to keep the normal performance of these networks. Nonetheless, in the trust evaluation process, dishonest nodes likely broadcast incorrect recommendations in the network. This decreases the accuracy of the trust value and affects the normal operation of the trust process. To solve this challenge, this paper presents an intelligent fuzzy logic-based trust system (IFTS) in UASNs. The proposed scheme employs a fuzzy trust mechanism to assess direct trust. To design this mechanism, energy evidence, data evidence, and communication evidence are considered as inputs in this fuzzy system, and direct trust is extracted as the fuzzy output. Energy evidence is obtained from the remaining energy and the energy change rate. Data evidence is obtained from the packet loss rate and data consistency, and communication evidence is calculated based on three link-related parameters, namely link reliability, link delay, and link stability. Likewise, recommendation trust depends on the recommendations offered by the recommenders. The trustor node evaluates each recommender and calculates its merit by using the root mean square (RMS) error and the trust value of the trustor relative to the recommender. Furthermore, IFTS computes indirect trust based on the trust chain, i.e., a set of recommender nodes. This trust chain is built using the greedy strategy based on the closest and most reliable recommender nodes. Further, IFTS uses a sliding time window for refreshing trust values. Finally, the simulation and evaluation process of IFTS is carried out in comparison with a recommendation management trust mechanism based on collaborative filtering and variable weight fuzzy algorithm (CFFTM), an adaptive trust model based on long short-term memory (LTrust), and a trust model based on cloud theory (TMC) under three attacks, namely bad/good mouthing attack, collusion attack, and hybrid attack, and its results are compared in terms of two criteria, i.e., diagnosis accuracy rate and false diagnosis rate. Hence, in the bad/good mouthing attack, IFTS improves the indirect trust level of honest nodes, accuracy, and the false diagnosis rate by 2.24%, 1.97%, and 12.68%, respectively. In the collusion attack, IFTS upgrades the indirect trust level of abnormal nodes, accuracy, and the false diagnosis rate by 7.2%, 1.17%, and 0.69%, respectively. In a hybrid attack, IFTS optimizes accuracy and the false diagnosis rate by 2.30% and 29.27%, respectively.