Struggling against selfishness and black hole attacks in MANETs

Since mobile ad hoc networks (MANETs) are infrastructureless and multi-hop by nature, transmitting packets from any node to another usually relies on services provided by intermediate nodes. This reliance introduces a new vulnerability; one node could launch a Black Hole DoS attack by participating in the routing protocol and including itself in routes, then simply dropping packets it receives to forward. Another motivation for dropping packets in self-organized MANETs is resource preservation. Some solutions for detecting and isolating packet droppers have been recently proposed, but almost all of them employ the promiscuous mode monitoring approach (watchdog (WD)) which suffers from many problems, especially when employing the power control technique. In this paper we propose a novel monitoring approach that overcomes some WD's shortcomings, and improves the efficiency in detection. To overcome false detections due to nodes mobility and channel conditions we propose a Bayesian technique for the judgment, allowing node redemption before judgment. Finally, we suggest a social-based approach for the detection approval and isolation of guilty nodes. We analyze our solution and asses its performance by simulation. The results illustrate a large improvement of our monitoring solution in detection versus the WD, and an efficiency through our judgment and isolation techniques as well. Copyright © 2007 John Wiley & Sons, Ltd.