In this paper, we study the issue of defending against adversarial localization in wireless sensor networks. Adversarial localization refers to attacks where an adversary attempts to disclose physical locations of sensors in the network. The adversary accomplishes this by physically moving in the network while eavesdropping on communication messages exchanged by sensors, and measuring raw physical properties of messages like Angle of Arrival, Signal Strength of the detected signal. In this

more »

... , we aim to defend sensor networks against such kinds of adversarial localization. The core challenge comes from the sensors performing two conflicting objectives simultaneously: localize the adversary, and hide from the adversary. The principle of our approach and the subsequent defense protocol is to allow sensors intelligently predict their own importance as a measure of these two conflicting requirements. Only a few important sensors will participate in any message exchanges. This ensures high degree of adversary localization, while also protecting location privacy of many sensors. Extensive simulations are conducted to demonstrate the performance of our protocol.