Optimal sensor scheduling and remote estimation over an additive noise channel

Estimation;Control of communication networks;Information theory and control

Gao, Xiaobin ; Basar ; Tamer

关键词: Estimation;    Control of communication networks;    Information theory and control;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/72772/Xiaobin_Gao.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

In the applications of wireless sensor networks, sensors are built to measure the state of the system of interest and send their measurements to a remote decision unit via wireless communication. Based on the messages received from the sensors, the decision unit estimates the state of the system and makes decisions. In this scenario, the quality of decision making strongly depends on the quality of state estimation. On the other hand, the sensors are constrained by limited power and cannot always communicate with the decision unit. As a consequence, a communication scheduling strategy and an estimation strategy should be designed for the sensors and the decision unit, respectively, such that the state estimation error is minimized under the communication constraints. In this thesis, we consider a sensor scheduling and remote estimation problem with one sensor and one estimator. The sensor makes a series of observations on the state of a source and then decides whether to transmit each one in the sequence to the estimator. The sensor is charged a cost for each transmission. The remote estimator generates real-time estimates on the state of the source based on the messages received from the sensor. The estimator is charged for estimation error. In contrast to prior work in the literature, we further assume that there is additive communication channel noise, which makes the problem more challenging. As a consequence of the presence of channel noise, the sensor needs to encode the message before transmitting it to the estimator. For some specific distributions of the underlying random variables, we obtain a person-by-person optimal solution to the problem of minimizing the expected value of the sum of communication cost and estimation cost over the time horizon, which is globally optimal in the asymptotic case. In a modified problem we show that our solution is locally optimal and a globally optimal solution exists.