Outlier Detection using Naive Bayes in Wireless Ad Hoc Networks

Nowadays, security is one of the most significant concerns when constructing a flexible and improvisational network. For a wireless ad hoc network, which is exposed to an open and cooperative environment, its vulnerability needs a more effective protection for the validation of information sharing, when compared to traditional networks. At present, trust gains extensive attention as it is regarded as a well-known distributed management method to perceive abnormal behavior of other nodes. In this paper, we emphasize the importance of node cooperation, especially for the sharing of trust information. Thereby, an outlier detection scheme is presented based on Naive Bayes algorithm, which is used to predict the reliability of trust information provided by other adjacent nodes. We examine the scheme based on our design criteria and attack models. Through the security analysis, Naive Bayes makes the trust-based outlier detection more suitable and reliable for distributed networks.