LiPaD: lightweight packet drop detection for ad hoc networks

We present a scalable, effective and practical approach for detection of packet-drop attacks in ad hoc networks. In this attack, a malicious network node chooses to selectively drop packets that are supposed to be forwarded, which results in adverse impact on application goodput and network stability. Briefly, our approach requires network nodes to report statistics on IP flow packets originated, received, or forwarded to neighbors. These statistics are analyzed and correlated to determine nodes suspected of dropping packets. Preliminary OPNet-based simulation results suggest that our approach is quite accurate in detecting single and multiple malicious packet-dropping nodes even in the presence of nonmalicious "natural" link-loss, and when malicious nodes are "lying" by reporting incorrect statistics. We are in the process of augmenting the simulation to better handle increased network traffic and node mobility, so as to determine the performance of the approach in more realistic scenarios.