Distributed guaranteed two-target tracking over heterogeneous sensor networks under bounded noises and adversarial attacks

Abstract This paper is concerned with the distributed guaranteed estimation for tracking two interacting mobile targets over a multi-sensor network in the presence of unknown-but-bounded noises and various adversarial attacks. First, a heterogeneous sensor network framework in terms of two distinct groups of sensors is employed to monitor the two targets. Each sensor in either group possesses different sensing, processing and communicating capabilities, thereby leading to distinct communication topologies among intra- and inter-group sensors. Second, a unified attack model is established to skillfully accommodate multiple adversarial attacks including node manipulation attacks and deception attacks. Third, two different groups of distributed consensus-based estimators are delicately constructed to deal with the network heterogeneity. Criteria for designing the desired estimators are then derived such that the true states of the two maneuvering targets are guaranteed to be enclosed by the calculated estimate ellipsoids at each time step regardless of the noises and attacks. Furthermore, two tractable optimization algorithms, in both single- and two-target tracking cases, are applied to recursively calculate the smallest possible ellipsoidal estimate sets. Finally, numerical verification of distributed two-vehicle tracking is carried out to demonstrate the effectiveness and applicability of the obtained results.