针对基于Intemet的实时机器人系统,提出了多尺度的时延预测算法;该算法是在线和实时的.基于网络结构分析了网络时延的主要组成及其时延特性.不同于传统的网络往返时延的预测,该算法预测的是能真实反映端对端数据包传递的单程时延,而且还提出了预测单向时延所需的时钟同步算法.理论分析和多点之间的网络实验验证了该算法的高效性.
This paper presents a multi-scale delay prediction algorithm for real-time telerobotic systems over Internet, and this algorithm is performed in an on-line and real-time manner. Based on the structure of Internet, the main composition of network delay and its characteristics are analyzed. This algorithm is designed to predict the one trip time delay, which can reflect the real delay of data packet from peer to peer, other than the traditional round trip time delay. In order to predict the one trip time precisely, a time synchronization algorithm is also proposed. The usefulness and effectiveness of the developed algorithm are validated by theoretical analysis and Internet experiments between multiple places.