基于预测的机会式网络编码  

Opportunistic network coding based on prediction

作　　者：刘外喜[1,2] 余顺争[2] 高鹰[1] 胡晓[1] 

机构地区：[1]广州大学电子信息工程系,广东广州510006 [2]中山大学电子通信工程系,广东广州510006

出　　处：《通信学报》2013年第4期36-46,共11页Journal on Communications

基　　金：国家自然科学基金资助项目广东联合基金重点项目(U0735002);国家自然科学基金资助项目(60970146;61202271);广东省自然科学基金资助项目(S2011040004068;S2012040007184)~~

摘　　要：针对理论网络编码在实际应用上的缺陷,以及现有机会式网络编码完全依赖于消息偷听导致应用场合的局限性,以完全不同于网络编码优化问题的思路,提出了基于预测的机会式网络编码的方法。其主要思想是:基于网络流量的自相似性,利用EMD(经验模式分解)和ARMA(自回归滑动平均)预测下一个报文的到达时间,综合计算编码时间、为了编码而等待的时间、传输时间等要素,从而决定是否编码。推导出了网络编码在不同情况下要实现吞吐量正增益可以等待的时间上界。仿真实验结果显示,在吞吐量上,提出的方法相对于理论网络编码平均提高15%左右,相对于当前的存储转发平均提高21%左右。在提高吞吐量的基础上,提出方法也可有效地降低网络的能量消耗。Since pure theory network coding had some drawbacks in practical network, however current opportunistic network coding fully depends on overheard information. Thus, in a radically different way from network coding optimi- zation, a novel opportunistic network coding scheme was proposed which was based on prediction （ONCP）. The main idea of ONCP was.. firstly, arrival time of next packet in a node was predicted for the self-similarity of network traffic; secondly, encoding time, waiting time for coding, transmission time and other factors were taken into account together to determine whether this packet was coded in this node. A theorem--maximum waiting time for was proved network cod- ing in order to obtain throughput gain. Simulation results show that ONCP improves throughput by 15% compared to pure theory network coding and by 21% compared to traditional store-forward. ONCP can also effectively reduce energy consumption while improving throughput.

关 键 词：机会式网络编码 预测 吞吐量 最长等待时间 经验模式分解 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]