多径路由可为互联网提供灵活的路由传输服务。然而，目前实际部署的多径路由技术仍存在明显不足，如多协议标签交换技术（MPLS）会增加封装负载且适用范围有限，策略路由（PBR）因为静态配置而效率较低。IP地址作为互联网报文固有元素，可蕴含身份、位置、拓扑等重要信息，却未被路由机制充分利用。本项目以源、目的IP地址为驱动，研究灵活、可控、可信的多径路由机制，并针对流量传输优化及可信控制等问题展开具体研究。研究内容包括：1）研究IP地址语义，构建IP地址驱动的多径路由设计模型；2）基于二维路由研究拥塞链路的快速规避方法，设计科学的分径流量遴选规则；3）研究多出口网络的外访流量控制，设计低代价二维转发路径构建方法及故障后快速路径切换机制；4）研究基于可信级别的多径传输，设计多关联拓扑最优路径的高效计算算法；5）研究转发引擎多径转发处理机制和表项存储优化，设计多径路由实验模型及验证环境。

Multi-path routing technology provides flexible transmission services for the Internet. The actually deployed multi-path routing mechanisms still have obvious shortcomings. For instance, Multi-Protocol Label Switching (MPLS) results in extra overhead of label encapsulation, which is only suitable for a few network scenarios. Besides, Policy Based Routing (PBR) is less efficient due to its static configuration. As an inherent field in IP packet header, IP address implies a lot of important information including identity, position, topology, etc. However, it is currently not fully utilized in IP routing. This project takes source & destination addresses as the driving elements to study flexible, controllable, and credible multi-path routing mechanisms and to study the traffic transmission optimization and trustworthy control in the Internet. The main work is illustrated as follows: 1) IP address semantics and the design model of the IP address driven multi-path routing; 2) Rapid avoidance method of the congested link based on the two dimension IP routing, and the scientific rules of selecting the bypass flows; 3) Outgoing traffic control in the multi-egress network, which includes the construction of two dimension forwarding path with low cost and fast path switching after the failure of a key link or node; 4) Multi-path transmission based on credible grades, and the efficient algorithm of calculating the shortest paths for the multiple associated topologies; 5) Multi-path forwarding processing mechanism and tables storage optimization, and experimental model and verification environment towards multi-path routing.