Localizing sensor networks in un-friendly environments

In this paper, we study the issue of defending against a wireless sensor network (WSN) that has been deployed by a malicious enemy agent in an area of interest to us. While there can be many approaches to defend against maliciously deployed WSNs, we propose the design of a localization centric approach. Specifically, the problem we address is: given an enemy deployed WSN in an area of interest to us, how can we determine locations of the sensors without co-operating with the sensors themselves during localization. In our approach, we employ a physically mobile agent called the localizer (e.g., a mobile robot) to move in the sensor network and detect raw sensor-to-sensor communication signals. However, the localizer has no information on the message content or the sensor id of any signal (possibly due to message encryption) since the sensors belong to an enemy agent. Based on estimating the angle of arrival and the received signal strength, we design a protocol for the localizer to determine sensor positions. The salient features of our protocol are efficient association of signals with sensors, and filtering many likely false locations over time. Sound theoretical analysis and extensive simulations are used to demonstrate the performance of our protocol from the perspective of localization accuracy.