MATF: a multi-attribute trust framework for MANETs

Abstract

To enhance the mobile ad hoc networks (MANETs) security, various trust-based security schemes have been proposed. However, in most of the trust-based security schemes, a node’s trust is computed based on a single trust attribute criteria, such as data forwarding. Using single trust attribute criteria may cause the bootstrapping problem, which refers to the time required by the trust-based scheme to build trust and reputation among nodes in the network. The bootstrapping problem in these schemes may provide more opportunities to misbehaving nodes to drop packets and remain undetected for longer time in the network. Moreover, using single trust attribute criteria does not effectively deal with the selective misbehavior by a smart malicious node. In this work, we propose a scheme that is based on the multi-attribute trust criteria to minimize the bootstrapping time, which ultimately improves the performance of the scheme in terms of high malicious node detection rate, low false-positive rate, and packet loss rate. The contributions of this paper are (a) identification of trust attributes along with the development of a comprehensive multi-attribute trust framework (MATF) using multiple watchdogs for malicious node identification and isolation, (b) formal modeling and verification of our proposed MATF using HLPN, SMT-Lib, and Z3 Solver, and (c) simulation-based validation and evaluation of the proposed trust framework in the context of optimized link state routing (OLSR) protocol against various security threats, such as message dropping, message modification, and link withholding attacks. The simulation results revealed that the proposed trust framework achieves about 98 % detection rate of malicious nodes with only 1–2 % false positives. Moreover, the proposed MATF has an improved packet delivery ratio as compared to the single attribute-based scheme.