SAT: an SVM-based Automated Trust Management System for Mobile Ad-hoc Networks

Abstract

Mobile Ad-hoc Networks (MANETs) are extremely vulnerable to a variety of misbehaviors because of their basic features, including lack of communication infrastructure, short transmission range, and dynamic network topology. To detect and mitigate those misbehaviors, many trust management schemes have been proposed for MANETs. Most rely on pre-defined weights to determine how each apparent misbehavior contributes to an overall measure of trustworthiness. The extremely dynamic nature of MANETs makes it difficult, however, to determine a set of weights that are appropriate for all contexts. We describe an automated trust management scheme for MANETs that uses machine learning to classify nodes as malicious. Our scheme is far more resilient to the context changes common in MANETs, such as those due to malicious nodes altering their misbehavior patterns over time or rapid changes in environmental factors, such as the motion speed and transmission range. We compare our scheme to existing approaches and present evaluation results obtained from simulation studies.