基于时延不对称网络的最小排队转发时延估计

针对精密时间协议(PTP)报文的不确定排队转发时延恶化PTP同步性能的问题,提出最小排队转发时延估计算法.通过对链路时延堆栈式比较,基于筛选出的未受阻塞的幸运报文进行时钟偏移估计,提高了时延不对称网络中的PTP同步性能.算法的边界条件由搜寻幸运报文的平均时间间隔决定,推导了该边界条件公式.测试验证结果表明:在主从链路时延不对称的环境下,相比于无优化情况,链路时延估计精度提高了约25ns,其稳定度提高了近2个数量级;PTP系统同步精度提高了1倍以上,其稳定度提高了近4倍.

Packet Delay Estimation with Minimum Queue Waiting in Asymmetric Network

To solve the problem of the deterioration in precision time protocol (PTP) synchronization performance caused by uncertainty queue waiting in packet transmission, this paper proposed a packet delay estimation algorithm with minimum queue waiting. Since the packet delay could be estimated with the lucky packets selected from stacks, the PTP synchronization performance would be improved. The boundary conditions of the proposed algorithm was decided by the mean time interval using the selection of “lucky packets” and its expression was deduced. Some meaningful tests indicate that in the asymmetry network, the accuracy of packet delay estimation increases almost 25ns with stability increasing almost 2 orders, and the accuracy of PTP synchronization increases more than one fold with stability increasing almost four folds, compared with no optimization.