Statistical en-route filtering of injected false data in sensor networks

Abstract

In a large-scale sensor network individual sensors are subject to security compromises. A compromised node can be used to inject bogus sensing reports. If undetected, these bogus reports would be forwarded to the data collection point (i.e., the sink). Such attacks by compromised nodes can result in not only false alarms but also the depletion of the finite amount of energy in a battery powered network. In this paper, we present a statistical en-route filtering (SEF) mechanism to detect and drop false reports during the forwarding process. Assuming that the same event can be detected by multiple sensors, in SEF each of the detecting sensors generates a keyed message authentication code (MAC) and multiple MACs are attached to the event report. As the report is forwarded, each node along the way verifies the correctness of the MAC's probabilistically and drops those with invalid MACs. SEF exploits the network scale to filter out false reports through collective decision-making by multiple detecting nodes and collective false detection by multiple forwarding nodes. We have evaluated SEF's feasibility and performance through analysis, simulation, and implementation. Our results show that SEF can be implemented efficiently in sensor nodes as small as Mica2. It can drop up to 70% of bogus reports injected by a compromised node within five hops, and reduce energy consumption by 65% or more in many cases.