Design And Implementation Of CQF Shaper Based On Linux In Time-Sensitive Networks

With the continuous upgrading of industrial digitization and informatization,industrial network technology supporting deep integration,interconnection and deterministic support capability has become an inevitable evolution direction.In order to solve the problems of heterogeneous information fusion and delay certainty between industrial Internet,time sensitive network is proposed.Among the shapers that provide delay guarantee for time sensitive networks,Time Aware Shaper(TAS)has precise gating time requirements for switch ports and high complexity in configuration.The end-to-end delay of Cyclic Queuing and Forwarding(CQF)is only related to hops and gating cycle.Therefore,the complete implementation of CQF mechanism has become a key issue worthy of being researched and solved.The IEEE 802.1Qch standard only defines the basic working principle of CQF.The specific implementation methods of CQF,especially the resource scheduling in CQF,belong to open problems.Up to now,there are few complete implementations.This thesis designs and implements a time sensitive network CQF shaper based on Linux.According to the current network state and under the limitation of bandwidth and end-to-end delay,this technology can independently plan the path and scheduling parameters,which are sent to the switches for configuration.Finally,the flow to forward can have the optimal scheduling strategy.The main work of this thesis includes:(1)The architecture of CQF shaper system is designed and implemented.Centralized user configuration in the control plane is used to collect and display information.Centralized network controller is used to run the scheduling algorithm and distribute the configuration information.The database is used as the information storage medium to establish the interactive channel between the control plane and the data plane.In the data plane,the switch based on Linux can carry out corresponding gating management according to the CQF configuration information issued.Then the classification and queue shaping of business flow is performed.So the scheduling mechanism of circular queue is completed,which ensures the deterministic transmission of business flow.This CQF shaper system has strong interactivity for users and can uniformly manage all equipment.It realize the accurate deployment of filter classification mechanism and gating configuration in the switch,which completely realize the CQF shaper mechanism.(2)CQF scheduling algorithm is designed and implemented.Based on IEEE 802.1Qch standard,the working mechanism of CQF is mathematically modeled.The parameters are created to express its components.the constraints of delay and bandwidth are established.The constraint design of key parameter time slot period T is carried out.Then,based on reinforcement learning,the CQF scheduling algorithm is designed for the mathematical model,which can balance the resource allocation under the dual requirements of bandwidth and end-to-end delay according to the current network state information.CQF algorithm can find the path that can reserve bandwidth resources for future business flows,so that more data flows can be accommodated in CQF scheduling.(3)In this thesis,the above CQF scheduling algorithm and CQF system implementation architecture are tested and verified.The test results show that the CQF scheduling algorithm has a good performance in the success rate of data flow scheduling.It is verified that this algorithm can give priority to the links with abundant bandwidth resources within the allowable delay range.It take a longer path within the limit,and improve the scheduling tolerance ability of business flow.In addition,the performance of CQF system implementation architecture is tested,and the end-to-end delay distribution before and after CQF scheduling is compared.The results show that the transmission certainty of service flow is greatly improved.the deterministic relationship between delay and hop number of forwarding path and scheduling cycle is satisfied.And the accurate scheduling of CQF mechanism is realized.