基于PTPd2的非对称链路时间偏差补偿方法
|
摘要
针对IEEE 1588v2主时钟到从时钟的时延和从时钟到主时钟的时延不相等的问题,构建了一种三层BP神经网络模型,将基于时间戳的主从时钟偏差计算结果输入该网络模型,对时间偏差进行补偿。实验结果表明:该方法能够将非对称链路的时钟同步精度由100μs提升至50μs,同时消除过大的抖动,从而更好地校正传输时延。
Aiming at the problem that the IEEE 1588 v2 master clock to slave clock delay and the slave clock to master clock delay are not equal, a three-layer BP neural network model is constructed, the calculation results of master-slave clock deviation based on timestamp are input into this network model to compensate the time deviation. The experiment results demonstrate that this model can improve the clock synchronization accuracy of asymmetric links from 100 μs to 50 μs, while eliminating excessive jitter, thus better correcting the transmission delay.
Aiming at the problem that the IEEE 1588 v2 master clock to slave clock delay and the slave clock to master clock delay are not equal, a three-layer BP neural network model is constructed, the calculation results of master-slave clock deviation based on timestamp are input into this network model to compensate the time deviation. The experiment results demonstrate that this model can improve the clock synchronization accuracy of asymmetric links from 100 μs to 50 μs, while eliminating excessive jitter, thus better correcting the transmission delay.
引文
[1]李培基,李卫,朱祥维,等.网络时间同步协议综述[J].计算机工程与应用,2019,55(3):30-38.
[2]梅军,徐迅,钱超,等.应用于配电网的时钟偏移估计的IEEE 1588改进方案[J].电力系统保护与控制,2016,44(9):83-89.
[3]张大帅,葛文萍,车全江,等.基于IEEE1588同步协议的本地时钟频率控制算法[J].计算机工程与设计,2016,37(2):308-312.
[4]IEEE Std.1588v2-2008,IEEE standard for a precision clock synchronization protocol for networked measurement and control systems[S].Piscataway:IEEE,2008:111-123.
[5]黄华明.IEEE1588V2路径延迟自动测量和非对称性补偿[J].光通信技术,2016,40(2):56-59.
[6]王志斌,孔亚楠,刘永成,等.遗传算法优化BP神经网络的大功率LED结温预测[J].光电子·激光,2014,25(7):1303-1309.
[7]PENG X,MEN Z,WANG X,et al.The ship motion prediction approach based on BP neural network to identify Volterra series kernels[C]//Control and Decision Conference(CCDC 2014),May 31-June 2,2014,Changsha,China.Piscataway:IEEE,2014.
[8]QIAN K,WANG T,TANG T,et al.A novel Local BP Neural Network model and application in parameter identification of power system[C]//Construction Conference(CCC 2014),July 28-30,2014,Nanjing,China.Piscataway:IEEE,2014.
[9]李依泽,陆超,王印峰,等.基于Kalman滤波与神经网络的高精度同步时钟算法[J].电网技术,2019,43(3):777-784.
[10]景振华.基于ARM9的B类LXI信号发生器的研究与实现[D].西安:西安电子科技大学,2013.
[11]苏宇,胡珩,张涛,等.基于PTPd的精准时钟同步技术的研究[J].计算机技术与发展,2016,26(1):175-180.
[12]WANG T,CHANG F.Application of PTPd to find the threshold of precision time synchronization[C]//System Science and Engineering(ICSSE 2011),June 8-10,2011,Macao,China.Piscataway:IEEE,2011.
[13]KOVACSHAZY T,FERENCZ B.Performance evaluation of PTPd,a IEEE 1588 implementation,on the x86 Linux platform for typical application scenarios[C]//Instrumentation and Measurement Technology Conference(I2MTC 2012),May 13-16,2012,Graz,Austria.Piscataway:IEEE,2012.
[2]梅军,徐迅,钱超,等.应用于配电网的时钟偏移估计的IEEE 1588改进方案[J].电力系统保护与控制,2016,44(9):83-89.
[3]张大帅,葛文萍,车全江,等.基于IEEE1588同步协议的本地时钟频率控制算法[J].计算机工程与设计,2016,37(2):308-312.
[4]IEEE Std.1588v2-2008,IEEE standard for a precision clock synchronization protocol for networked measurement and control systems[S].Piscataway:IEEE,2008:111-123.
[5]黄华明.IEEE1588V2路径延迟自动测量和非对称性补偿[J].光通信技术,2016,40(2):56-59.
[6]王志斌,孔亚楠,刘永成,等.遗传算法优化BP神经网络的大功率LED结温预测[J].光电子·激光,2014,25(7):1303-1309.
[7]PENG X,MEN Z,WANG X,et al.The ship motion prediction approach based on BP neural network to identify Volterra series kernels[C]//Control and Decision Conference(CCDC 2014),May 31-June 2,2014,Changsha,China.Piscataway:IEEE,2014.
[8]QIAN K,WANG T,TANG T,et al.A novel Local BP Neural Network model and application in parameter identification of power system[C]//Construction Conference(CCC 2014),July 28-30,2014,Nanjing,China.Piscataway:IEEE,2014.
[9]李依泽,陆超,王印峰,等.基于Kalman滤波与神经网络的高精度同步时钟算法[J].电网技术,2019,43(3):777-784.
[10]景振华.基于ARM9的B类LXI信号发生器的研究与实现[D].西安:西安电子科技大学,2013.
[11]苏宇,胡珩,张涛,等.基于PTPd的精准时钟同步技术的研究[J].计算机技术与发展,2016,26(1):175-180.
[12]WANG T,CHANG F.Application of PTPd to find the threshold of precision time synchronization[C]//System Science and Engineering(ICSSE 2011),June 8-10,2011,Macao,China.Piscataway:IEEE,2011.
[13]KOVACSHAZY T,FERENCZ B.Performance evaluation of PTPd,a IEEE 1588 implementation,on the x86 Linux platform for typical application scenarios[C]//Instrumentation and Measurement Technology Conference(I2MTC 2012),May 13-16,2012,Graz,Austria.Piscataway:IEEE,2012.