VANDER: Efficient Cooperative Watchdog Monitoring for Lossy Wireless Network Coding

Abstract

Network coding achieves multicast network capacity by allowing intermediate nodes to mix information in packets. However, due to the mixing operation, network coding is vulnerable to pollution attacks. Different from conventional cryptographic solutions, watchdog-based solutions, which have been recently proposed for network coding, rely on trusted nodes to monitor and verify behaviors of transmission nodes. However, the impact of lossy wireless communications has not been considered. Since false alarm and misdetection depend on the loss rate, these false results can happen with high probability when packet loss happens. In this paper, we propose VANDER, which is a novel cooperative watchdog scheme in heterogeneous wireless networks, where multiple watchdogs collaborate and efficiently detect pollution attacks in a lossy wireless environment. The novelty of our approach is that, when lossy overhearing happens, watchdogs work cooperatively to share the packet information, where no extra overhead is introduced to normal transmission nodes, and rather than retransmitting all lost packets among watchdogs, watchdogs use randomly generated Vandermonde hashes to detect corrupted packets. Moreover, VANDER is capable of detecting successive colluded adversaries. In addition to the low false alarm and misdetection probabilities, VANDER also achieves low computational complexity and communication overhead. Numerical experiments are provided to support the theoretical analysis of VANDER.