Constrained remote state estimation problems in the context of Cyber-physical systems are studied in this thesis. Communication resources constraints such as insufficient network bandwidth and sensor battery power shortage often impose challenges on the state estimation performance. To make the best use of limited communication resources, the concept of controlled communication is introduced. By using a sensor scheduling approach, we can obtain a desirable tradeoff between the estimation quality and the communication cost.

We mainly focus on two classes of constrained state estimation problems: the communication-rate constrained estimation problem over reliable networks, and the energy constrained estimation problem over unreliable networks. Though this research area was extensively ex...[ Read more ]

Constrained remote state estimation problems in the context of Cyber-physical systems are studied in this thesis. Communication resources constraints such as insufficient network bandwidth and sensor battery power shortage often impose challenges on the state estimation performance. To make the best use of limited communication resources, the concept of controlled communication is introduced. By using a sensor scheduling approach, we can obtain a desirable tradeoff between the estimation quality and the communication cost.

We mainly focus on two classes of constrained state estimation problems: the communication-rate constrained estimation problem over reliable networks, and the energy constrained estimation problem over unreliable networks. Though this research area was extensively explored in the last decade, there are still open questions to solve. For example, the existing solutions to the communication-rate constrained estimation problem are mostly complicated and practically useless, while we tackle the nonlinear estimation problem therein by introducing a stochastic event-triggering mechanism. We also improve the tradeoff for the energy constrained estimation problem by using the feedback information of the channel condition. Briefly speaking, we propose some novel efficient sensor schedules by utilizing the realtime information during transmission. The scheduling pattern is affected by some time-varying parameters and is thus stochastic. The theoretical and experimental analysis shows the efficiency and computational simplicity of the proposed stochastic sensor schedules.