DoS Attack Allocation Against Remote State Estimation Over Packet-Dropping Networks

We consider the security problem of remote state estimation in cyber-physical systems. The sensor sends data under round-robin protocol to the remote side while denial-of-service (DoS) attacker could interfere this process. Due to the energy-constrained DoS attack, it needs to decide whether to launch attacks to the wireless channel. In this work, two different cases for multi-sensor running Kalman filter are studied, i.e., sensors have limited or sufficient computational abilities. Furthermore, to capture the unreliability of the communication network, the occurrence of packet loss in the absence of attack is considered. Based on the evolution process of estimation error covariance, we first propose the sufficient condition to guarantee the optimality of the attack allocation strategy which maximally degrades the system performance. Then, the optimal attack sequence is given in a closed-form and shown to be monotonic with respect to time. Finally, a simulation example is given to illustrate the effectiveness of our proposed attack strategy.