面向可靠应用层组播的分层树网模型及差错控制研究

The reliable application layer multicast is a new technique which has a vast application prospect. The key problem of the technique is how to perform effective error control over the basic data distribution of the application layer multicast (ALM). In ALM, dynamic member host also acts the role of data transfer node, which brings some features (e.g., dynamic structure, error correlation and hierarchical difference) to the ALM delivery tree. The above features make the error control over ALM become a complex and intractable problem. To address the above problem, this project attempts to propose a new forwarding/recovering model (i.e., the hierarchical tree-mesh model) for reliable application layer multicast. In the new model, the multicast data is forwarded along the low-level ALM delivery tree, while the data recovery is performed based on the high-level mesh structure, which is generated in terms of the low-level tree. The proposed model can provide better structural basis for the error control of reliable ALM, without reducing the high-effective forwarding capability of the tree structure. This project will further study the error control of reliable ALM based on the proposed model and the features of the ALM delivery tree, using some reasonable techniques (e.g., distributed cooperation), with multiple optimization objectives (mainly including high scalability, low bandwidth consumption and high capability of adapting the dynamics). The expected research achievements can provide new theoretical and technical supports for the error control of reliable application layer multicast.

新兴的可靠应用层组播技术具有广阔的应用前景，其需要解决的核心问题是如何在应用层组播基本数据分发基础上实施有效的差错控制。在应用层组播中，由于动态成员主机同时担当数据转发角色，应用层组播转发树具有动态性、错误相关性和层次差异性等特征，这些特征使得可靠应用层组播差错控制成为一个难以有效解决的复杂问题。针对上述情况，本项目将提出一种新的可靠应用层组播转发/恢复模型即分层树网模型。在拟提出的模型中，数据转发由底层应用层组播转发树承担，数据恢复则在由转发树推演而成的高层网状结构上进行，从而在保留转发树高效转发能力的同时，为相应差错控制提供更加合理的结构基础。进一步地，本项目将充分结合拟提出的新模型和应用层组播特征，通过分布协同等技术手段，以易扩展、低带宽消耗和高动态适应能力为主要性能优化目标，对可靠应用层组播差错控制关键问题展开研究。项目研究成果可为可靠应用层组播差错控制提供新的理论与技术支撑。

可靠应用层组播技术核心问题是如何在应用层组播基本数据转发基础上实施有效的差错控制。在应用层组播中，由于动态成员主机同时担当数据转发角色，应用层组播转发树具有动态性、错误相关性和层次差异性等特征，这些特征使得可靠应用层组播差错控制成为一个难以有效解决的复杂问题。本项目对基于转发树的应用层组播特征进行了分析，并提出了分层的可靠应用层组播差错控制模型。在提出的模型中，数据转发由底层应用层组播转发树承担，数据恢复则在独立的逻辑结构上进行，从而在保留转发树高效转发能力的同时，为相应的差错控制提供了更加合理的结构基础。项目组研究了基于网状结构的数据恢复结构，利用多源增强数据恢复的可靠性。此外，也研究了基于环状结构的数据恢复结构。在上述工作基础上，项目组对应用层组播分层数据恢复、基于云虚拟机的可靠应用层组播及其差错控制、应用层组播树构建和优化方法等展开了有效的研究。本项目也对软件定义网络环境下的组播可靠传输问题及差错控制进行了研究。