Distributed set-membership estimation against mixed measurement and communication attacks

The distributed set-membership state estimation is addressed for linear multi-sensor systems subject to denial-of-service (DoS) and false data injection (FDI) attacks on measurements and communication channels. A unified hybrid attack model is proposed, where DoS attacks and FDI attacks are characterised by Bernoulli-distributed random sequences with dynamically adjustable probabilities to reflect realistic adversarial behaviours. Then, a novel distributed set-membership filtering framework is developed, which guarantees that the actual state of the system remains within a sequence of ellipsoidal sets, and sufficient conditions for the feasibility of the designed filter are derived. Furthermore, a retransmission mechanism and joint optimisation scheme are introduced to minimise the trace of the estimation error ellipsoid, thereby enhancing both robustness and estimation accuracy. Finally, simulation results validate the performance of the developed approach.