基于多元时间序列的无线网络跨层设计研究

Research on Wireless Network Cross Layer Design Based on Multivariate Time Series

摘要: 针对无线网络传输过程中，容易出现某些节点负载压力过大的问题，利用跨层设计思想，结合多元时间序列理论，提出一种基于时间序列的节点负载预处理机制。利用跨层设计方法监测节点内分组投递率和网络吞吐量的变化情况，并以此构建模糊时间序列预测模型，实现对节点负载压力的动态预测。仿真实验表明，基于分组投递率和网络吞吐量的模糊时间序列可以实现节点负载压力的有效预测，有助于避免无线网络中“热点”节点的形成，进而达到提高无线网络传输质量的目的。

Abstract: Specific for overloading problems in wireless network transmission process, combined cross-layer concept and multivariate time series’ theory, a preprocessing mechanism for node load based on multivariate time series is proposed. In consideration of the cross-layer design, the packet delivery ratio and network throughput are monitored to establish the multivariate time series forecasting model, which will realize the dynamic prediction for nodes’ load pressure. Simulation results show that the multivariate time series prediction model for packet delivery ratio and network throughput can effectively predict the nodes’ load pressure, which can effectively avoid the formulation of congested nodes and thus to improve the transmission quality of the wireless networks.