业务流预测在网络编码平台的实现及验证

Implementation and evaluation of network traffic prediction in network coding

网络编码由于可以利用无线信道的广播特性带来的空间冗余条件,在无线自组织网络中的应用显示了巨大的潜力。但TCP业务的机会网络编码在无线自组织网络中获得的增益却很小。基于这一问题,提出对TCP业务流具有的自相似性进行FARIMA模型建模,进行数据包到达时间的预测,从而为路由节点根据网络状态进行正确的编码等待决定提供依据,仿真结果验证了TCP业务流的自相似性,预测算法具有良好的预测结果。文中详细介绍了预测算法在COPE网络编码的NS2平台中的具体实现流程。

With the spatial redundancy provided by the broadcasting nature of wireless channel, network coding shows great potential in ad hoc network application. However, the throughput gain of TCP traffic with network coding keeps low. This paper proposes the model for TCP network traffic based on its self- similarity using FARIMA model and then predicts the arriving time interval of data packets for routing nodes, enabling them to make right coding decisions based on the current network situation. The simulation result proves the self-similarity of TCP traffic and the predicting algorithm shows the good performance. In the paper the applying process of the predicting algorithm in COPE on NS2 is introduced in detail.