| 摘要: |
| 在通信网络的设计中，使用基于流量预测的网络规划已成为LTE发展的必然趋势。与
地面网络不同，卫星网络由于受资源受限和拓扑时变的不利影响，其流量预测算法必须能兼
顾精度和效率，这令传统的地面网络预测方法已不再适用。为了解决以上问题，提出了一种
新的基于小波回声状态网络的流量预测算法，该算法通过小波多尺度分解的信号处理方法屏
蔽了网络流量的噪声，而后结合了无反馈的回声状态网络联合进行预测。仿真证明，新算法
相比传统算法能大幅提升网络流量的预测精度和运行效率，为卫星网络的流量规划提供了
强有力的决策支持。 |
| 关键词: 卫星通信网 流量规划 小波回声状态网络 多尺度分解 |
| DOI：10.3969/j.issn.1001-893x.2013.07.002 |
|
| 基金项目:国家高技术研究发展计划(863计划)项目(2012AA01A507) |
|
| A new traffic flow prediction algorithm for satellite communication network |
| QIN Hong-xiang,YANG Fei |
| () |
| Abstract: |
| In the design of a communication network, it has become an inevitable tr
end with LET development to use traffic flow prediction method for network plann
ing. Due to suffer from the constrained resource and changeable topology, the tr
affic flow prediction algorithm of the satellite networks which is different fro
m terrestrial networks, should fully take into account the accuracy and efficien
cy, so the traditional prediction methods for terrestrial networks are no longer
fit for application. In order to resolve this problem,a new traffic flow pre
diction algorithm based on wavelet and echo state networks is proposed in this
paper. The new algorithm uses the signal processing method based on multi
-scale decomposition of wavelet to shield the noise of the network traffic, and it comb
ines with the non-feedback echo state network to predict. The simulation result
s show that the new algorithm can greatly improve the prediction accuracy and run
ning efficiency of network traffic compared with traditional algorithms, so it p
rovides a more scientific decision support for satellite communication network t
raffic planning. |
| Key words: satellite communication network flow planning wavelet echo state network multi-scale decomposition |
《电讯技术》编辑部 