Performance Analysis And Optimization Of Multi-path TCP For Large-scale AGV

In the national development strategy of the new infrastructure,the Industrial Internet of Things(Io T)has significant prospects as an important industry for information infrastructure and has a Huge challenge.Smart warehouse is a typical project in industrial Io T that can reduce costs and increase efficiency.However,it has a series of problems of perception,decision making,communication and execution.This paper focuses on the communication problems in smart warehousing.When the scale of smart warehouse increases,a large number of Automated Guided Vehicles(AGV)need to be used for the work.The communication problem between large scale AGVs and control servers is an unprecedented challenge.The problem of large scale AGVs disconnection problem has occurred in the world's second largest intelligent unmanned warehouse.In this paper,the following research is done on this problem.In this paper,we analyze the AGV disconnection problem in a large-scale AGV control network in a real-world smart unmanned warehouse.We find that it is the roaming problem of the wireless network card that is the main cause of AGV disconnection.We propose a MPTCP-based large-scale AGV control network solution to solve this problem.We design and test the MPTCP-based large-scale AGV control network solution.Finally,we fully apply the MPTCP-based large-scale AGV control network solution to an intelligent unmanned warehouse.The experimental results show that the MPTCP-based large-scale AGV control network solution solves the large-scale AGV disconnection problem.In this paper,we measure a MPTCP-based large-scale AGV control network.We designed and implemented a large-scale AGV control network measurement system.We use this system to measure the performance of the MPTCP-based largescale AGV control network.We use a decision tree model to statistically analyze the measured data.We found that the main factors affecting the delay of large-scale AGVs are the motion condition of the AGVs and the roaming condition of the AGV's wireless NICs.In this paper,we point out the causes of AGV dropout problem and AGV pause problem in large-scale AGV control networks based on multi-path TCP,and we optimize and improve the problem.We propose the Keep Full Mesh path management algorithm to solve the AGV dropout problem.The algorithm monitors the state of sub-streams in real time and ensures the simultaneous existence of multiple substreams.We propose the Probe Jitter Redundant scheduling algorithm to solve the AGV pause problem.The algorithm probes on infrequent sub-streams and switches to redundant mode when link jitter is detected or timeout retransmission occurs.We provide detailed evaluations the Keep Full Mesh path management algorithm and the Probe Jitter Redundant scheduling algorithm by replaying the link data in a large-scale AGV network.The test results show that the Keep Full Mesh path management algorithm and the Probe Jitter Redundant scheduling algorithm can significantly reduce the AGV disconnection problem and the AGV pause problem.