代间网络编码的若干关键技术研究

发布时间：2018-06-30 10:49

本文选题：MGM + 编码参数　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：相比于传统网络的存储转发模式，通过网络编码可以得到更好的网络传输性能，它允许我们以最大流-最小割定理得出的最大容量传输信息。但是由于传统网络编码存在解码计算量大、实时响应能力差、缓存空间需求大等缺点，因此出现了代内网络编码。然而当网络丢包率较大时，部分代的分组丢失导致接收不到足够的分组，以至于无法恢复出该代的原始信息，于是提出了多代混合(MGM,Multi-Generation Mixing)网络编码——“代”间网络编码，将文件分成多个混合集，每个混合集包含若干代，同一个混合集中位置编号大的代总是和编号小的代混合编码。本文对MGM网络编码的基本原理和算法进行研究，取得了以下关键成果： (1)首次针对发送端在发送分组之前如何确定MGM网络编码参数(混合集容量，“代”集容量，冗余率)这一问题，定义了数据有效投递率、探讨了在解码成功率达到99%以上数据有效投递率最优的准则下的参数取值。 (2)本文利用泊松模型，首次从时延的角度分析了Internet网络中MGM的编/解码成功率。提出了编/解码等待时延的具体算法、证明了编码节点的分组输出依然服从泊松过程、通过仿真验证了当等待时延在合理区间内时，Internet网络的网络编/解码成功率接近于1。 (3)针对“代”间网络编码方案解码计算量及解码时延较大的问题，提出了一种新颖的“代”间网络编码方案，即邻“代”混合(AGM,Adjacent-Generation Mixing)网络编码，将混合集中的每一代与它的前一代进行混合编码，编码过程与它的相邻两代有关，与混合集中的其他代无关。理论分析和仿真结果表明AGM编码方案相比于MGM以及已有的MGM改进方案能够大量减少解码计算量和解码时延，同时并保持与MGM相当的解码成功概率。
[Abstract]:Compared with the traditional network storage and forwarding mode, the network coding can get better network transmission performance, which allows us to obtain the maximum capacity of transmission information based on the maximum flow-minimum cut theorem. However, the traditional network coding has many disadvantages, such as large decoding computation, poor real-time response and large cache space, so the intra-generation network coding appears. However, when the packet loss rate of the network is high, the packet loss of some generations results in insufficient packets being received, so that the original information of the generation can not be recovered. Therefore, the MGMU Multi-Generation Mixing network coding, "intergenerational" network coding, is proposed. The file is divided into multiple mixed sets, each of which contains several generations, the generation with large number in the same hybrid set is always and the generation with small number is always mixed. In this paper, the basic principles and algorithms of MGM network coding are studied, and the following key achievements are obtained: (1) for the first time, how to determine the MGM network coding parameters (mixed set capacity, "generation" set capacity) before sending packets is the first time for the sender to determine the MGM network coding parameters. (2) in this paper, we define the effective delivery rate of data, and discuss the parameter value under the criterion of the optimal effective delivery rate of data over 99% of the decoding success rate. (2) in this paper, we use Poisson's model. The success rate of encoding / decoding of MGM in Internet network is analyzed for the first time from the point of view of delay. A specific algorithm of coding / decoding waiting delay is proposed, and it is proved that the packet output of the coding node still follows the Poisson process. The simulation results show that the network coding / decoding success rate of Internet network is close to 1 when the waiting delay is within a reasonable range. (3) the problem of large decoding computation and decoding delay for the "intergenerational" network coding scheme is discussed. In this paper, a novel "generation" network coding scheme is proposed, that is, the adjacent "generation" hybrid network coding, in which each generation in the hybrid set is mixed with the previous generation, and the coding process is related to its adjacent two generations. It has nothing to do with other generations in a mixed set. The theoretical analysis and simulation results show that the AGM coding scheme can greatly reduce the decoding computation and decoding delay compared with MGM and the existing MGM improved scheme, while maintaining the same probability of success as MGM.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.01

【参考文献】