基于RFieldbus的Token-Passing协议实时性能建模

Real-time Performance Modelling of Token-Passing based on RFieldbus

RFieldbus 无线现场总线的 MAC 协议 Token Passing 可看成是一个多台多队列排队系统。在分析了 TTR(令牌目的循环时间)对 RFieldbus 中 3 个优先级队列的服务时间的动态限制后,多台单队列排队模型 RF-P 被提出。对从站的轮巡交换数据任务被转化为RF-P 中的令牌内走步时间,原有的高、低级别两个队列则化成了两个具有不同 Poisson 到达率的 M/G/1/∞服务台。利用排队论中的准守恒 PCL 定律,在将服务时间限制转换为数量限制后,求出了等待响应时间的数学解析式。与试验结果的比较表明,在负载不大于 60%时其偏差度约在 10-20%间。

The Token-Passing protocol in the RFieldbus (Radio-Fieldbus which comprise two heterogeneous transmission media: wireless- and wired-segment) system acts as a multi-server-multi-queue system. A multi-server-single-queue RF-P model was built after the analyzing of the dynamic boundary of Token-visiting time of Token-Passing protocol in RFieldbus. In this model, the polling-time on the slaves was transformed into an internal token walking time when TTR is large enough, and the rest two queues were transformed into two M/G/1/ infinity systems with two different Poisson-arrival rates, so that the complexities of the time properties in RFieldbus can be depicted easily by 7 different parameters. With the help of Pseudo-Conversation-Law and the K-limited scheduling an equation to evaluate the average waiting time in RF-P model was derived. The experiment results show that the error of this equation is round 10-20% when system load is lower than 60%.