| Because deterministic network technology can take into account both high bandwidth and high certainty at the same time,it is gradually becoming a research hotspot in hard real-time fields such as industrial control,aerospace,and autonomous driving.The deterministic network technology represented by time-sensitive networks adds time characteristics to the standard Ethernet,and provides strict deterministic guarantees by pre-planning the transmission of critical traffic.In the above process,the end system is the terminal for generating and receiving critical traffic,and the certainty of its processing process has an important impact on the transmission effect of critical traffic.In actual operation,not only tasks related to specific applications need to be run on the end system,but also tasks related to network management such as time synchronization.Therefore,in order to improve the real-time response and provide multi-task support,the deterministic end system usually adopts a real-time operating system as a task running platform,and an embedded processor provides instruction processing support.For periodic or time-sensitive tasks,the real-time operating system is usually based on the task scheduling mechanism of time slice interruption,allowing the time slice interruption to interrupt the task execution process to ensure the real-time response of high-priority tasks.However,the above mechanism faces the following challenges when applied to deterministic end systems: 1)There is uncertainty in task scheduling time;2)There is uncertainty in task execution time.These jitters will seriously affect the end system’s certainty of critical traffic processing.In response to the above challenges,this paper proposes a task scheduling optimization technology for deterministic end systems.The main work and innovations of this paper are as follows:1.The deterministic-oriented task scheduling software optimization technology STDS is proposed.On the basis of combing the implementation mechanism of the task scheduler of the current end system,it analyzes the reasons of uncertainty and scheduling delay,proposes a task planning model that can meet constraints,and designs a deterministic task scheduling mechanism.STDS reduces the jitter in the task scheduling process and task execution process,and reduces the task scheduling delay.2.Proposed the deterministic task scheduling hardware optimization technology HTDS.Aiming at the uncertainty of the task execution process,a hardware scheduler is designed on the existing embedded processor,and the task planning table and the software scheduling decision process are hardware offloaded to realize the parallelism of the processor and task scheduling.Through interrupt shielding and other technologies to ensure the continuous execution of time-determined tasks,reduce task execution time jitter,and reduce task scheduling and execution process delay.3.Completed the prototype design of STDS and HTDS,and tested and evaluated the prototype system.Experimental results show that STDS can reduce the time overhead of critical task scheduling to 11.8% and the time overhead of task interrupt processing to4.7%.HTDS can completely eliminate the jitter in the task execution process,and further reduce the task scheduling time by 30% on the basis of STDS.In summary,in order to solve the task scheduling problem of the existing deterministic end system,the STDS and HTDS mechanisms proposed in this paper can significantly improve the determinism of task scheduling and execution process,and achieve higher scheduling performance.For the deterministic end system Design has important theoretical significance and practical value. |
