Network Time Analysis And Application Research

With the advent of the era that everything is connected,accurate,reliable,and highly available network time is the basis for the normal operation of all network application scenarios.Using the network to transmit time information has attracted more and more attention.Network time synchronization can meet the requirements of most applications' time synchronization without additional synchronization equipment.It takes the advantages of low cost,wide use,and convenience.In addition,with the diversification of network application scenarios,the demand for reliability,high precision,and so on for network time synchronization is motivated.The purpose of this article is to build a Internet time service monitoring system to continuously monitor Internet time services worldwide.The collected data is used to analyze the performance of the monitored time services,study the factors that affect the performance of Internet time services,and evaluate Internet time services to provide a certain reference for the user sides.At the same time,according to the current and future application requirements for time synchronization,we explore the situations that network time service meets the application requirements and the network factors that affect the application time synchronization.And,we provide solutions to improve the accuracy of network time synchronization.This article focuses on the performance analysis of network time services and the application of high-precision network time synchronization.The details are as follows:(1)An Internet time service monitoring system is proposed.This system can provide accurate reference time from GNSS and monitor hundreds of Internet time services worldwide.Then,We proposed an Internet time service performance analysis model,and three metrics for evaluating time service performance:availability,stability and accuracy.In order to obtain the anomaly of the Internet time service and the data set after removing the anomaly,we design an anomaly detection algorithm to ensure the accuracy of the analysis results.Finally,the performance of the monitored Internet time service was analyzed based on the service performance metrics.Experimental results show that 56 TSPs' availability are more than 95%.It denotes that the number of link nodes has no effect on the performance of the Internet time service.Moreover,the accuracy is positively related to the standard deviation index in stability.(2)We propose a high-precision time synchronization scheme for financial transactions.After analyzing the need for time synchronization and network factors that affect time synchronization in financial scenario,a software-defined time synchronization scheme is proposed.This scheme is based on a software-defined method,using time synchronization management and control separation.Then,in order to eliminate the jitter of time synchronization,a PID control algorithm based on BP neural network is proposed,and this algorithm can be embedded in the controller of the proposed scheme.Then,We set up the experimental environment to conduct comparative experiments.First,we compare the traditional time synchronization method with the proposed scheme.It is concluded that the time synchronization accuracy can be improved by about 5 times compared with the traditional time synchronization scheme,and the proposed software-defined time synchronization scheme is further verified that it meets and exceeds the current time synchronization requirements of financial transactions.Then,we select the digital PID control algorithm and Kalman filter algorithm as the comparison algorithm to verify that the performance of PID control algorithms based on back-propagation neural networks is better than the other two algorithms,and can obviously improve the time synchronization accuracy and maintain the data characteristics.