Optimal Transmission Scheduling for Remote State Estimation Under Active Eavesdropping-Based DoS Attacks
研究了存在主动窃听并实施DoS攻击的网络中，传感器如何通过联合设计功率控制和调度决策来最小化自身能耗和估计误差，同时最大化窃听者的估计误差，并提出了基于Q学习的在线近似最优策略。
In this article, the optimal transmission scheduling problem for remote state estimation over SINR-based network channel is studied, in the presence of an active attacker who is able to implement DoS attacks to jam the network based on the eavesdropping information. An intelligent sensor is used to send the local state estimates to a remote estimator, and by co-designing the power control and the scheduling decision such that the sensor can decide whether to transmit and what power to use for communication, a coupling transmission strategy is provided. To minimize the energy consumption and the known estimation error covariance (EEC) of the remote estimator for the sensor, while maximizing the unknown eavesdropping EEC, the co-design scheduling issue is modeled as a modified MDP by applying a Monte Carlo method based on a belief state probability distribution. A Clipped HetUpSoft <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Q</i>-learning algorithm is designed to achieve the approximate optimal strategy online. Finally, simulation results are provided to validate the effectiveness of the developed approaches.