AVB网络流量整形帧模型端到端延迟计算

音视频桥接技术(AVB)以传统以太网为基础,通过基于信用量的整形(CBS)算法进行流量整形,提供了音视频流确定性传输的基础保障。在CBS流模型的延迟分析计算基础上,考虑了AVB网络中数据帧实际排队和调度的离散特性,提出了一种改进的CBS帧模型来计算音视频流量的排队传输延迟,并推导了帧模型下最坏延迟计算公式。采用典型的车载网络应用场景对2种模型进行演算验证,结果表明:同流模型相比,CBS帧模型中得到的端到端延迟更小,显示出更好的紧性评估效果,同时这种紧性评估在针对低优先级流量的延迟分析时,差异会更加显著。

End-to-end traffic latency computation using frame shaping model in AVB network

Audio video bridging (AVB) Ethernet is an available Ethernet-based solution for audio and video stream transmissions within embedded networking context,in which credit-based shaping (CBS) algorithm is defined to guarantee the end-to-end latency performance according to stream models.When considering the discrete characteristics of frame queuing and scheduling behavior in an actual networking scenario,an improved frame model CBS was proposed to replace the stream model CBS algorithm.According to this new model,the worst-case transmission latency was deduced by eliminating non-schedulable traffic in the network.A representative in-vehicle application scenario was established to verify the end-to-end latency performance of the frame model CBS.The results indicate that the frame model CBS can obtain a better end-to-end latency evaluation than the stream model CBS.Especially,the tightness of frame model CBS is clearer for the lower priority AVB traffics within the whole AVB networks.