Fast state estimation under sensor attacks: A sensor categorization approach

Abstract

In a sensor network, some sensors usually provide the same or equivalent measurement information, which is not taken into account by the existing secure state estimation methods against sparse sensor attacks such that the computational efficiency of these methods needs to be further improved. In this paper, by considering the observation equivalence of sensor measurement, a concept of analytic sensor types is introduced based on the equivalence class to develop a fast state estimation algorithm. By verifying the similarity degree of measurement data of a sensor type, the attack location information can be extracted to exclude some mismatching search candidates, without loss of estimation correctness. This confirms high speed performance of the proposed algorithm, since the number of sensor types is usually far less than the number of sensors.