On-line Transmission Scheduling For Wireless Remote State Estimation Under Unkown Channel

Under the in-depth integration of computing technology,control technology,and communication technology,Cyber Physics Systems(CPS)have emerged.CPS combine the physical dimension and the information dimension,which have important applications in such fields as industrial production,national defense military,and intelligent transportation.Wireless remote state estimation,which faces the challenges of limited sensor resources and unreliable communication network,is one of the most important technologies in CPS.Existing researches have proposed network resource scheduling to improve the performance of remote state estimation.Nevertheless,most of these researches are based on assumed channel packet loss models(i.i.d,Markov,etc.).However,in practical applications,the state of the wireless channel is difficult to judge in advance because it is affected by a variety of complex factors,especially under complex working conditions.Therefore,the on-line transmission scheduling of wireless sensors is proposed.The main contributions of this work as follows: 1.The background and an overview of the current status of research of CPS and remote estimation are introduced.The deficiency of current research works is pointed out.2.The Kalman filter and the improved Kalman filter for unreliable communication are introduced.We also introduce the basic theory that used for model predictive control.3.An event-based retransmission adjustment scheduling for wireless sensor in single remote state estimation system of CPS under unknown channel is proposed.According to the arrival condition of the data packet,an adaptive method that can adjust the retransmission times of the packet to be sent in its transmission time slot is designed,so as to improve the transmission reliability.The experimental results show that the proposed scheduling can reduce the packet loss rate significantly and keep the covariance of the remote state error at a relatively low level compared with the average pre-allocated transmission scheduling.4.Considering the energy limitation,a wireless sensor transmission energy control method based on model predictive control(MPC)under unknown channel is proposed.In this method,the packet transmission time slot can be transmitted with different energy levels.Based on the prediction of the future time-domain channel quality by using deep learning,the optimal time-domain transmission level selection sequence is solved by using MPC.The experimental results show that the proposed scheduling can reduce the packet loss numbers by 80% and keep the covariance of the remote state error at a relatively low level compared with the average pre-allocated energy scheduling.Finally,the works are summarized and we look forward to the future research.