Maintaining functional module integrity in sensor networks

Security is one of the top concerns when sensors are deployed in hostile environments for military applications. Attacks on sensor nodes can be categorized into two types: attacks through radio signals to alter the functionalities of sensors and physical attacks; which may completely damage sensors. We show that the former is easier to launch and more disguised, thus more hazardous. The layered framework proposed in this paper is specially designed to defend against the first type of attacks. Different from precious reputation-based approaches or data mining technologies which usually incur large communication and computation overhead, our solution only involves localized computation which is less intensive, thus can be implemented on-board with little burden on current MICA2 platforms. Besides identifying malicious nodes, our solution can also recover the normal functionalities of sensors, which have been altered by malicious radio signals