Research And Implementation Of Mixed Flows On-demand Scheduling Technology In Time-sensitive Networks

With the gradual advancement of intelligent manufacturing in China,the construction of deep integrated industry networks has been a new momentum nowadays.As an emerging technology,time-sensitive networking(TSN)is applied to industrial converged communication scenarios,which has become a hot spot.However,due to the complicated traffic types and requirements in time-sensitive networks,traffic scheduling has been an urgent problem to be solved.This thesis designs and implements a scheduling technology for mixed flows in time-sensitive networks to solve the fine-grained traffic scheduling problems.This technology uses differentiated methods for different types of flows to plan forwarding paths,transmission timeslots and parameters of credit-based shaper to meet the corresponding delay and bandwidth requirements.The main work of this thesis includes:(1)The time-sensitive network prototype system is designed and implemented.This thesis utilizes the switches and hosts that combine the Linux OS and commercial network adaptors to build up a hardware environment,and develops software to realize the functions of control and switching in TSN.Meanwhile,this thesis raises a classification method for industrial network flows.According to various characteristics,flows are divided into three types: time-triggered(TT)flow,audio and video(AVB)flow,and besteffort(BE)flow.Every class employs different shapers and scheduling methods to meet the corresponding transmission requirements.(2)This thesis designs and realizes a capacity analysis algorithm for time-sensitive networking.This algorithm is used to schedule and allocate resources for TT flows,AVB flows and BE flows.According to the transmission characteristics of different types of flows,different methods are used to analyze and calculate the forwarding paths and shaper configurations that meet the requirements.In detail,the algorithm applies the Satisfiability Modulo Theories(SMT)to solve the forwarding paths and the transmission timeslots for TT flows.In addition to that,an analytical latency model for AVB flows based on Network Calculus is constructed,from which the calculated end-to-end latencies are acquired.The algorithm schedules the forwarding path and credit-based shaper parameters of AVB flows based on calculated end-to-end latencies.What's more,the forwarding paths of BE flows can be planned by the forwarding path planning algorithm.This algorithm solves the scheduling problems of mixed flows in TSN and greatly improves the scheduling ability in complex scenarios compared with similar algorithms.(3)This thesis tests and analyzes the time-sensitive network prototype system and the proposed capacity analysis algorithm.The time-sensitive network system is tested functionally according to the work flow of networking,registration,reconfiguration and operation.Compared to the performance of flows before and after reconfiguration,the results verify the Qo S guarantee ability of the system model.Besides,the performance of the capacity analysis algorithm is tested separately by the transmission timeslots scheduling algorithm and the analytical latency model for AVB flows.The results reveal that the transmission timeslots scheduling algorithm has favourable performance in running time and success rate,and the analytical latency model for AVB flows has high reliability,effectiveness and accuracy as well.