Research Of Communication Strategy In Real-time Control System Based On Age Of Information

In recent years,with the greatly development of the 5th Generation Mobile Networks(5G)technology,the Industrial Internet has gradually become a hot research topic.In-dustrial Internet aims to realize the ubiquitous interconnection of people,machines,and things.The largest challenge of Industrial Internet is how to transmit information with low latency and high reliability.Also,the proliferation of Industrial Internet application scenarios(e.g.,customized production,smart factories,etc.)has significantly boost the demand for fresh status update information to guarantee the good real-time performance.To achieve these requirements,it is necessary to optimize the design of the communica-tion system and control system.The traditional design method optimize communication and control system independently,while the control performance under this design obvi-ously does not meet the requirements of Industrial Internet.Thus we need to jointly design communication and control system to improve control performance.This thesis jointly consider the communication system and the real-time control sys-tem.Based on a new metric which quantifies the freshness of information called age of information,two communication strategies are proposed to improve the real-time perfor-mance of the control system.The main contribution of this thesis are as follows:Firstly,this thesis proposes a Hybrid Automatic Repeat Request(HARQ)strategy based on AoI.Compared with the traditional HARQ strategy which issue a retransmis-sion request immediately after packet loss,our proposed strategy transmit new state data packets or issue a retransmission request based on the current state and the impact of deci-sion for the overall performance of the system instead of directly choosing retransmission.On the basis of this strategy,we construct the Markov decision process of the packet trans-mission process,and use the DQN method of reinforcement learning to find the solution.The simulation results show that compared to the communication strategy that does not adopt HARQ and the communication strategy that adopts traditional HARQ,our proposed strategy can significantly improve the performance of control.Secondly,this thesis proposes a packet management strategy at the receiver based on AoI.Compared with the traditional design from the perspective of the source to minimize AoI,we found that actively discarding some low-value information at the receiver can also reduce AoI.We built an optimization model to minimize AoI,calculated the average AoI of the receiver through an iterative algorithm,and calculated the active packet loss threshold through a simulated annealing algorithm.The simulation results show that the proposed strategy can significantly reduce the average AoI of the receiver compared with the traditional strategy.Thirdly,in this thesis,an experimental platform test is carried out on the receiver package management strategy.By designing robotic arm as an actuator and depth camera as a sensor,at the same time,relying on the universal software radio peripheral(USRP)customized communication system,the ”remote real-time control of robotic arm” experi-mental platform is built in this thesis.We take the real-time error of the motion trajectory(Tracking Error)of the robotic arm and the experimenter's hand as an index to measure the control performance.Through experimental tests,we found that the strategy we proposed can reduce the Tracking Error.