Energy-Efficient Co-Design of Power Scheduling for State Estimation Over a Stochastic Delayed Network
研究了延迟网络中的调度问题，提出混合随机模型同时考虑延迟和丢包，给出离线最优解和基于绝对阈值的在线调度策略，并证明在线策略全局优于离线策略。
In this article, the scheduling problem over a delayed network is investigated. Different from the existing literature, we propose a hybrid stochastic model which incorporates both delay and packet loss. Based on this proposed model, an energy-efficient co-design problem is considered under power-limited scenarios. A global optimal offline solution is presented explicitly. Aided by arriving information feedback, we propose an online schedule based on an absolute threshold. The optimality of our proposed schemes and thresholds selection is theoretically analyzed. Meanwhile, we have extended the conclusions of the single-system case to the multisystem case. Optimal strategies from both offline and online perspectives are provided with rigorous proof, respectively. The stability condition is given for the estimator to converge in an infinite-time horizon. The comparisons between offline and online strategies are proved the global superiority of the online strategy to the offline one. Numerical simulations have validated the correctness and superiority of our proposed algorithms.