基于往返时延的网络拓扑推断

Network topology inference based on round trip time

为了减少拓扑推断中采用单向性能参数需要多个节点合作的限制,提出了一种基于往返时延的拓扑推断算法,设计了网络拓扑推断中的往返时延测量方法,基于往返时延的拓扑推断不需要时钟同步及目标节点的配合。从理论分析了基于往返时延推断网络拓扑结构的可行性和正确性,并通过NS2进行了仿真实验。仿真结果表明,基于往返时延的推断算法能够较准确地推断网络的拓扑结构,与基于单向性能参数的拓扑推断算法相比,基于往返时延的拓扑推断算法受到的限制较少。

In order to reduce the limits of topology inference based on one direction performance characteristics, a topology inference algorithm based on round trip time was proposed and a measurement method was designed in topology inference. Topology inference based on round trip time did not need cooperation and time synchronization between source node and destination nodes. The feasibility and correctness of topology inference algorithm based on round trip time were analyzed theoretically. The algorithm was validated through simulations of NS2. Results of simulation indicate that the topology inference algorithm based on round trip time can infer network topology correctly, and the topology inference algorithm based on round trip time has less limits than the topology inference algorithm based on one direction performance characteristics.