Clock Synchronization And Stability Analysis Based On Clock Drift Rate

The problem of Clock Synchronization is a classical issue in distributed operating system and core technology in distributed computing. As the development of computing technologies and network technologies rapidly, the new content and broad range are added in this research area of clock synchronization. The aim of this thesis is to present a new strategy of fitting offset to adjust time in order to achieve the effective synchronization between slave nod and reference node. This strategy adopts a continuous time-stamp communication model and a model of clock drift rate in asynchronous network environments. The result proves that it not only can overcome the impact of network delay and jitter on clock synchronization effectively, but also can obviously improve the stability of clock synchronization on client.There are two kinds of clock synchronization tasks: external synchronization and internal synchronization. The key difference of these two implementation methods is the presence of a standard time basis. If a time basis exists, we usually use the external synchronization policy to synchronization all clocks. In this study, the external synchronization policy is applied.Nevertheless, during the process of the external synchronization, the clock synchronization between different nodes confront with many difficulties. Firstly, due to the variations of transmission delay and other uncertain factors in the network, there will be haphazard larger clock offsets between slave node and reference node. Therefore, each computer cannot have an instantaneous time value to synchronize clocks. Secondly, even if all clocks could be started at the same real time, they would not remain synchronized because of drifting rates. Besides, their drift rate can be changed due to temperature variations or aging. The difference between two hardware clocks can thus change as time passes. Finally, the most important key point in the research of clock synchronization is how to enhance the stability of clock synchronization on client effectively.Based on the statements above, the research of this dissertation used continuous timed-stamp communication model to create statistical mathematical models of clock running precision difference. The application of continuous time intervals can achieve better precision than that of instantaneous time intervals, because this model was created by a lot of time-stamps accepted by this node. The error or drop of a little time stamps could not affect the precision of this model.Based on the continuous timed-stamp communication model, this study adopts the local time of slave node and the arrival timestamps, from the reference server, to create offset trend equation, and a mathematic model of clock drift rate is created by a method of linear fit. From this mathematic model, the clock drift rate of the slave node can be acquired. Then, this study uses this drift rate to adjust time and analyze stability.Based on the theory above, client software of clock synchronization was developed by this study. It not only can dynamic display the trend graph of roundtrip delay ,offset and offset trend equation at real time, but also can adjust time on the strategy of fitting offset or instantaneous offset. The result shows that fitting offset strategy not only can avoid the influence from the haphazard larger clock offset effectively, but also can enhance the stability of clock synchronization on client and clock running precision in a further step. In the meantime, this method also has a better adaptability of clock synchronization for a worse network transmission environment. Therefore, it will have a brilliant future.The chief objective in the thesis is to acquire better theoretical understanding of clock synchronization theory and method. A practical solution is discussed by simulation experimentations. Many of these tools, middleware and applications on grid require an identical clock system. In fact accurate global clock synchronization is invaluable for many applications -grid computing, distributed computing and applications.