摘要:为了从根本上解决现有互联网存在的可扩展性、移动性和安全性等方面的问题，全新的未来互联网体系结构得到了广泛研究.其中，命名数据网络（named data networking，简称NDN）利用网内缓存和多路转发实现了基于层次化名字的高效数据传输，从根本上解决了现有互联网所面临的问题.内容的层次化名字具有数量庞大、结构复杂等特点，现有的基于IP的路由转发机制无法直接应用于NDN网络，需要有针对性地研究高效的层次化名字路由机制，保证海量网络内容的正常路由转发.路由聚合是缩减网络路由规模的主要措施.不同于现有的面向本地NDN路由表查表过程的优化，路由聚合需要全网协同处理，在不同网络节点上不断对聚合路由进行聚合.这对聚合路由标识和聚合路由可用性评估提出了诸多要求.为此，研究并提出了针对层次化名字路由的聚合机制，包括两个方面的工作：（1）构建了一种全新的计数布隆过滤器——堆叠布隆过滤器，该过滤器支持多过滤器合并，用于压缩表示被聚合路由名字；（2）给出了一种动态路由聚合机制，在保证NDN网络路由转发准确性的同时，缩小全网路由规模，最大程度地优化了路由转发效率.在真实网络拓扑上构建了仿真平台，经过实验验证，该路由聚合机制以可控的少量冗余转发为代价，有效地压缩了全网路由规模，提升了全网路由转发效率，保证了海量在线内容的高效路由转发，为NDN网络投入实际部署提供了前提.



Abstract:TCP/IP-based Internet is faced with severe challenges in scalability, mobility, security, and controllability, which makes it necessary to research the clean-slate architecture for Internet. Named Data Networking (NDN) leverages in-network caching and multipath transmission to support data name-based packet transmission, and conquers the aforementioned challenges faced by the conventional Internet. Considering the huge number and the significant diversity of hierarchical names used in NDN, the existing IP-based routing cannot be directly applied, and efficient hierarchical name-based routing mechanisms are desired to gurantee stable data transmission. Route aggregation is the core mechanism for FIB size reduction in the TCP/IP-based Internet. Hierarchical name-based route aggregation leverages router cooperation over the whole network to aggregate routes recursively, different from the existing researches for name-based route lookup optimization on a single router. To achieve effective hierarchical name-based route aggregation, identifier generation is focused on in route aggregation and utility evaluation of the aggregated routes. Ultimately, a hierarchical name-based route aggregation scheme is proposed, including two parts:(1) A novel combinable counting Bloom filter for representing aggregated route names compactly, namely, Compounded Counting Bloom Filter (CCBF); (2) A dynamic route aggregation scheme over the whole network for aggregating routes recursively according to their utility. To evaluate the performance of the proposed route aggregation scheme, extensive simulations are performed based on real network topology. Evaluation results show that the proposed scheme can significantly reduce FIB size and improve FIB lookup efficiency at the cost of small number of false positive lookup results. This work provides a foundation for practical NDN deployment.



PDF全文下载地址:

http://jos.org.cn/jos/article/pdf/5572