On the pitfalls of high-throughput multicast metrics in adversarial wireless mesh networks

Recent work in multicast routing for wireless mesh networks has focused on metrics that estimate link quality to maximize throughput. Nodes must collaborate In order to compute the path metric and forward data. The assumption that all nodes are honest and behave correctly during metric computation, propagation, and aggregation, as well as during data forwarding, leads to unexpected consequences In adversarial networks where compromised nodes act maliciously. In this work we identify novel attacks against high-throughput multicast protocols in wireless mesh networks. The attacks exploit the local estimation and global aggregation of the metric to allow attackers to attract a large amount of traffic. We show that these attacks are very effective against multicast protocols based on high-throughput metrics. This leads us to conclude that aggressive path selection is a double-edged sword: it maximizes throughput, but in the absence of protection mechanisms it also increases attack effectiveness. Our approach to mitigate the identified attacks combines measurement-based detection and accusationbased reaction techniques. The solution also accommodates transient network variations and is resilient against attempts to exploit the defense mechanism itself. We demonstrate the attacks and our defense using ODMRP, a representative multicast protocol for wireless mesh networks, and SPP, an adaptation of the well-known ETX unicast metric to the multicast setting.