Adaptive network time synchronization

这项研究和开发项目是与北电网络合作进行的。该项目是分析和模拟时间同步的性质，在一个庞大而复杂的网络互连的计算机和无线基站。长期目标是开发方法，以高精度同步计算机和无线基站的大型网络。时钟同步的重要性在于使系统能够实现非常高的数据速率。为了实现精确的同步，必须考虑本地时钟频率、时钟偏置和不同传输延迟的微小差异。这些频率差异表现为时钟漂移。因此，我们需要不断估计各个时钟漂移和偏差，并考虑这些差异。还将有一些连接到全球定位系统接收器，将测量全球定位系统时钟时间。随着连接的时钟和支持GPS的基站数量的增加，我们将能够改善时钟同步。长期目标是将网络的同步量化为网络拓扑的函数。 网络拓扑会影响整个网络的传输延迟。我们将使用各种系统识别方法和学习算法来整合整个网络中的所有时间戳测量，以实现高度的同步精度。

This research and development project is being carried out in collaboration with Nortel Networks. The project is to analyze and model the nature of time synchronization across a large and complex network of interconnected computers and wireless base stations. The long term objective is to develop methods to synchronize a large network of computers and wireless base stations to a high degree of accuracy. The importance of clock synchronization is so that the system can achieve very high data rates. To achieve accurate synchronization one must account for small differences in local clock frequency, clock bias and varying transmission delays. These frequency differences manifest themselves as clock drift. Therefore, we need to continuously estimate individual clock drift and bias and account for these discrepancies. There will also be a number of connections to GPS receivers that will measure GPS clock time. As the number of connected clocks and GPS enabled base stations increase, we will be able to improve the clock synchronization. The long range goal will be to quantify the synchronization of the network as a function of the network topology. Network topology will affect transmission delays across the network. We will use various methods of system identification and learning algorithms to integrate all the time stamp measurements across the network to achieve a high degree of accuracy in synchronization.