Research On Traffic Scheduling Mechanism In Time Sensitive Networking

With the empowerment and expansion of mobile communication technology in vertical field,in addition to supporting ultra-reliable and low-latency communication services,a transmission network that can provide bounded delay and jitter is also crucial to the further development of 5G.On the one hand,the traditional wired network technology can only provide a delay of tens of milliseconds.On the other hand,deterministic solutions such as fieldbus technology and industrial Ethernet technology,which have been applied to certain extent at present,have problems such as high cost and unable to support the transmission of different performance traffic at the same time.Therefore,in order to meet the increasing demand for real-time transmission,Time Sensitive Networking(TSN)overcomes the limitations of previous technologies to provide deterministic quality of service for the next-generation 5G transmission network and support heterogeneous network communication.This paper focuses on the time sensitive networking traffic scheduling mechanism based on the 5G fronthaul environment,and analyzes the performance of the synchronous scheduling mechanism and the asynchronous scheduling mechanism respectively,and discusses the impact of the two on traffic transmission under different circumstances,so as to provide reference for the selection and design of time sensitive fronthaul system.Firstly,based on the time aware shaping mechanism proposed in the IEEE 802.1Qbv standard,a multi-objective gate control list implementation method using improved genetic algorithm and improved tabu search algorithm is designed to solve the static scheduling problem in offline mode.In order to prove the feasibility of the implementation method and the efficiency of the hybrid genetic tabu algorithm,it is verified on the OMNeT++discrete event simulation tool:the simulation results show that the obtained gate control list can meet the scheduling requirements of various types of fronthaul flows,and the hybrid genetic tabu algorithm functions better when solving speed and results in contrast to the modified genetic algorithm and the genetic algorithmsimulated annealing algorithm.In addition,based on the asynchronous traffic shaping mechanism proposed in the IEEE 802.1 Qcr standard,a switch parameter configuration method based on the core idea of software defined networking and reinforcement learning algorithm is designed to solve the dynamic scheduling problem in online mode.Firstly,the asynchronous traffic shaping mechanism and network centralized control module are implemented on OMNeT++,and then the reinforcement learning algorithm is used to calculate the scheduling parameters of the asynchronous traffic shaping mechanism in this centralized module.Finally,combined with the simulation results,the scheduling performance,the influence of shaping algorithms and parameters,and the necessity of improving the reinforcement learning algorithm of iterative efficiency of ascension are analyzed.