Attack-aware sliding mode control for 2-D Markov jump systems under important-data-based DoS attacks

This paper investigates output-feedback sliding mode security control for two-dimensional (2-D) Roesser Markov jump systems subject to important-data-based denial-of-service (DoS) attacks. A global-index-based 2-D jumping description is adopted to eliminate mode ambiguity along the two independent directions, so that a unique mode evolution is assigned to each diagonal layer. An attack-aware signal reconstruction mechanism is then developed from the defender's perspective, where a conservative proxy of the attacker's triggering threshold is inferred only from successfully received measurements. Based on the reconstructed indicator, an asynchronous output-feedback sliding mode controller is designed. Sufficient conditions are established to guarantee mean-square exponential stability and practical reachability of a bounded sliding region under bounded disturbances and targeted data losses. These conditions are further converted into tractable linear matrix inequalities by suitable variable substitutions. Simulations on a switched 2-D example and a Darboux-type model show that IDB attacks cause significantly more severe performance degradation than conventional DoS patterns, while the proposed control scheme effectively mitigates their impact.