Research On Time Synchronization Of High Latency Networks

High latency networks are those communications networks with large propagation delay, like under water communications network and deep space communications network. The former network has a low propagation rate which lead to large propagation delay, the latter network has a extremely long distance between nodes which also lead to large propagation delay. The high latency networks have large propagation delay and high frequency in link switching which causes inefficient of the application of the upper transport protocol. It's very important to build reliable and efficient upper-layer protocols in high latency networks.Time synchronization technique is very essential for high latency network, it should has a high-precision performance in time synchronization with low energy consumption. There are some protocols for high latency networks including TSHL, Tri-Message and so on. It's common for these protocols to synchronizing time by compensate the clock time using clock frequency drift and clock offset computed in the protocols.It is known that the clock frequency drift and clock offset variables are short-way-stable in a stable network, so we propose a smooth time synchronization protocol based on Tri-Message protocol. In this protocol, first, we do a serial time synchronization using Tri-Message and get a serial data of clock frequency drift, then make a smooth weighted computation on these variables, second, we transmit the synchronization frequency to W-times lower, then do the time synchronization periodically without the synchronization signals based on the variation of the clock frequency drift and clock offset. The proposed protocol is emulated in the OMNET++ platform and the results show that, when W=3, the data rate consumed is almost 50% less than Tri-Message.In some cases, the propagation delay is changing in some situations like mobile nodes. For this problem, we proposed a dynamic tracking synchronization protocol for the high latency networks with mobile nodes by computing the changing rate of propagation delay, and then estimated the propagation delay in time synchronization.