【作者】 张婧垚；

【导师】 樊平毅；

【作者基本信息】 清华大学 ， 信息与通信工程， 2010， 博士

【摘要】 网络编码是21世纪以来信息技术领域的一项重大成果。它打破了传统路由器只能简单地存储转发这一常规。通过将多个数据包组合在一起,并利用数据包之间的相关性来解码,网络编码巧妙解决了有向网络中组播最大流等经典的理论难题,大大提高了网络容量和资源的利用效率,并在优化网络管理和提高网络的安全性等诸多方面都具有重要价值。无线信道的广播特性使得网络编码在无线网络中有很大的应用潜力。作为一个新兴的研究领域,无线网络编码面临着无线终端的移动性、无线信道的复杂多变和终端资源受限等许多挑战,有很多关键问题需要解决。本文立足研究无线网络编码的若干关键理论和技术问题,完善现有的理论框架,利用网络编码思想来改善现有无线网络的性能,并为其实用化打下基础。在基础理论上,建立了一种无线网络的随机图模型,并给出基于无线网络编码的组播最大流路由算法,编码节点选择方法及优化的输出边容量分配策略,继而给出无线网络编码的完整实现框架。在此基础上,从理论上研究了端到端和组播的最大流容量分布,并量化分析无线网络编码的增益和实现复杂度。本文还研究了无向图中的网络编码,首次发现了信息流碰撞现象。此外针对网络编码特有的排队现象和等待延迟,本文从理论上量化分析了延时与传输效率等因素的关系,并提出相应的优化调度策略。在工程方面,本文将网络编码思想与传统的交叉层技术结合,以改善网络性能。第一项工作是基于网络编码的无线信号干扰消除技术,该技术利用已知的数据包内容,在无需附加同步和功率控制的条件下可有效消除信号干扰,从而增强网络的空间复用性,提高传输效率。另一项工作提出一种乘积码形式的分布式网络纠错码,可显著改善组播的误码性能。最后,为了让相关技术具有实用性,本文还探讨了网络编码在应用中需解决的实际问题,包括差错控制与重传、延时和QoS保证、及与现有系统的兼容等等,并给出相关的解决方案。基于无线局域网协议,设计并实现了无线网络编码硬件实验平台,解决方案的可行性通过实验得以验证。更多还原

【Abstract】 Network Coding is a momentous achievement of information technology in the 21st century. It breaks the convention that a router can only simply store and forward messages. Through combining multiple packets together and using the correlation among packets for decoding, network coding solves some classical theoretical problems such as the maximum multicast flow in directed networks in a very smart way, greatly increases the network capacity and the utilizing efficiency of resources, and has remarkable significance in enhancing the robustness of network management and data security. The broadcast nature of wireless channel endows network coding with extraordinary merit and potential in wireless networks. As a newly emerging technology, wireless network coding is faced with many challenges such as the mobility of wireless node, the variance of channel and limited resources of wireless equipments, and there are a lot of key questions to answer. This work aims at solving some key theoretical and technical problems of wireless network coding, completing the current theory framework, and using the idea of network coding to improve the performance of wireless networks, which lays the foundation for its practice.For the basic theory, the random graph model will be established for wireless networks, and put forward the strategies of maximum flow multicast routing, coding node selection as well as an optimized capacity allocation scheme on output links based on network coding, and further propose the whole framework of wireless network coding. Based on these works, we will study the distribution of the maximum flow for both end-to-end and multicast communication theoretically, and analyze the benefit and complexity of wireless network coding quantitatively. Network Coding in undirected networks will also be studied, with the original discovery on the collision phenomenon of information flows. Moreover, the relation between delay and efficiency will be investigated to the queuing and waiting delay that is peculiar to network coding, and optimized scheduling algorithms will be given.For engineering, we combine network coding with traditional cross-layer techniques to improve the network performance. The first job is an interference eliminating technique based on network coding. Through the content of packets that is known to the receiver, this technique can effectively eliminate the interference without synchronization and power control, thus increase the space reusing and the transmission efficiency is increased. The other job is distributed network error correction code in the form of product code, which can reduce the bit error rate of multicast.Finally, to make the concerning techniques practical, we will discuss the issues to deal with when network coding is implemented, such as error control, retransmission, delay, QoS guarantee and its compatibility with current systems, and present the corresponding solutions. The feasibility of these solutions is proved by the hardware experimental bed of wireless network coding designed in this work.更多还原