Two-Target Tracking Over Heterogenous Sensor Networks Under Deception Attacks

Abstract

This paper addresses the problem of two-target tracking over a heterogenous sensor network under deception attacks. To track the corresponding targets, the spatially distributed sensors are classified into two groups, and the sensors in each group are capable of exchanging measurement information only with their neighboring sensors in accordance with some prescribed interaction topologies. In the presence of deception attacks, the measurement received by each sensor suffers deliberate modification and thus the tracking performance of the two targets may be degraded or even disrupted. First, a heterogenous distributed estimation scheme based on the two distinct groups of sensors is developed to deal with the simultaneous effects of the unknown but bounded process noises as well as the physically constrained deception attacks. Second, criteria for designing the desired distributed estimators and the weights of interacting information links among the inter- and intra-group sensors are derived. It is shown that the true states of the two moving targets are guaranteed to be enclosed by two groups of estimate ellipsoidal sets at each time step regardless of process noises and deception attacks. Third, an optimization problem is proposed to minimize the obtained ellipsoids, aiming to provide optimal tracking performance. Finally, an illustrative example is given to demonstrate the effectiveness of the proposed target tracking method.