An attack-resilient distributed extended Kalman consensus filtering algorithm with applications to multi-UAV tracking problems

This study investigates how the events of deception attacks are distributed during the fusion of multi-sensor nonlinear systems. First, a deception attack with limited energy (DALE) is introduced under the framework of distributed extended Kalman consensus filtering (DEKCF). Next, a hypothesis testing-based mechanism to detect the abnormal data generated by DALE, in the presence of the error term caused by the linearization of the nonlinear system, is established. Once the DALE is detected, a new rectification strategy can be triggered to recalibrate the abnormal data, restoring it to its normal state. Then, an attack-resilient DEKCF (AR-DEKCF) algorithm is proposed, and its fusion estimation errors are demonstrated to satisfy the mean square exponential boundedness performance, under appropriate conditions. Finally, the effectiveness of the AR-DEKCF algorithm is confirmed through simulations involving multi-unmanned aerial vehicle (multi-UAV) tracking problems.