无线自组织网络中高能效的网络编码组播机制研究

Energy-efficiency is the key concern for the large-scale application of network coding based multicast in wireless Ad Hoc networks. This project investigates energy-efficient network coding based multicast schemes, which try to reduce the total transmission times of the nodes who take part in the multicast without scarifying other performance gains such as throughput, etc. First, we will utilize Markov chain to model the dynamic transmission procedure of the segmented random linear network coding, and analyze the energy-efficiency under some network characteristics, like broadcast channel and lossy link. We believe that these stuides can provide a theoretic guide for the design and analysis of the energy-efficient methods. Then, we will characterize the need for encoded packets of a node and its ability to forward encoded packets based on fuzzy mathematic theory. Guided by the idea of on-demand, we will build some optimization models and explore possible solutions. Specifically, we consider the selection of encoding nodes based on the match between the need for encoded packets and the ability to forward encoded packets, and the selection of encoding frequency of the nodes who are responsible for saving cry baby nodes. These studies will lay the foundation for energy-efficient performance. Finally, we will design low-cost feedback scheme with the help of packet piggyback and packet combination. Following this study, we will address the issues of the efficient and dynamic segment switch and parallel transmision of multiple segments with inter-segment cooperation to further improve the energy-effiency. It is supposed that our research results can not only enrich the theory and technique of green communications and network coding, but also give effective solutions to guarantee the QoS of multicast in wireless Ad Hoc networks.

能效性能是网络编码组播在无线自组织网络中能否大规模应用的关键，本项目研究高能效的无线自组网网络编码组播机制，力争在不损失吞吐量等性能增益的前提下，减少参与网络编码组播传输的节点的总通信收发次数来提高能效。首先，建立基于Markov的批次随机线性网络编码组播动态传输模型并结合广播信道、链路丢包等特征对能效性能展开分析，为高能效协议与方法的设计分析提供理论指导；接着，采用模糊数学相关理论表征节点对编码分组的需求及其编码转发能力，以按需为思路，建立基于需求与能力匹配的编码转发节点选择与Cry Baby拯救节点的编码转发速率选择等优化模型并求解，奠定高能效的基础；最后，设计基于捎带、合并技术的低耗反馈机制，由此对高效动态批次切换与基于批次间协同的多批次并行传输策略进行研究，进一步保障高能效。项目成果可以丰富绿色通信与网络编码等新兴学科的理论与技术，为无线自组网组播业务的QoS保证提供解决方案。

经过4年的研究，项目取得了丰硕的成果，超额完成项目指标。共发表重要学术论文32篇，其中SCI检索16篇，EI检索16篇，申请国家发明专利10项，已授权3项，相关成果获得省部级科技进步三等奖2项，培养博士生3名，硕士生8名。项目取得的主要研究成果包括：.1，提出了适用于单跳认知无线网络认知用户的网络编码组播机制。首次在考虑可用传输时长不确定性的情况下研究了基于网络编码的组播问题。针对授权信道可用传输时长序列服从的分布已知的情况，提出利用置信区间估计对可用传输时长的期望值进行预测；针对授权信道可用传输时长序列服从的分布未知的情况，提出引入多臂老虎机理论对当前可用传输时长进行动态决策。.2，提出了适用于多跳认知无线网络认知用户的网络编码组播机制。首次在考虑信道可用带宽不确定性的情况下研究了基于网络编码的组播问题。在充分考虑信道可用带宽不确定性的情况下将问题建模成一个机会受限问题。鉴于该问题较难解，利用Bernstein近似将问题转化成一个凸优化问题，保证了转化后的问题的解相对原问题解的可行性和较容易求解，并利用Lagrange松弛提出了一种分布式的机制。.3，提出了适用于移动认知无线网络认知用户的网络编码组播机制。首次在考虑可用信道数目不确定性的情况下研究了基于网络编码的组播问题。将问题建模为在满足即时组播交付率约束和链路传输约束下最小化总传输耗费的优化问题。利用离散时间Markov模型刻画信道的可用性和节点的移动性，并进一步将问题近似为一个混合整数规划问题，提出了一个具有频谱感知能力的基于耗费的分布式组播传输机制。.4，提出了基于滑动窗口的自适应网络编码协议。采用自适应滑动窗口机制，根据反馈信息动态调整数据处理粒度，从而获得了更好的通信环境能力；在此基础上，利用滑动窗口限制了网络中传输的未知分组数目，从而避免了目的节点处解码矩阵无限增大，并改进了滑动窗口的反馈方式，在不影响滑动窗口机制正常工作的前提下，减少了反馈开销。