基于地理定位的互联网拓扑测量关键技术研究

The topology of the Internet is critical in understanding and precisely simulating the Internet, and it is also very important in optimizing large-scaled Internet applications. However, existing Internet topology measurement approaches lead to incomplete topological results, and incur redundant probes, which degrade the probing efficiency considerably. In addition, there lack methodologies for accurately mapping the measurement results to the PoP-level topologies. In this project, we focus on key technical issues in Internet measurement and topology mapping, and seek to make contributions in the following aspects: 1) We seek to propose a novel algorithm for retrieving probing targets from BGP routing tables, and design a task scheduling mechanism that enables multiple vantage points to collaboratively probe the Internet, so as to improve the efficiency of the Internet measurement as well as the completeness of the topological results. 2) We seek to propose a methodology for mapping the PoP-level Internet topologies; we improve the accuracy of the resulting topology by accurately geolocating IP addresses and PoP nodes on the topology. 3) We seek to develop a PoP-level routing model based on the analysis of the PoP-level topologies, and propose an algorithm for inferring the PoP path between two arbitrary IP addresses; the proposed model and path inference algorithm will serve as a foundation for people to better understand and manage Internet's routing behaviors. 4) We will design and implement an Internet topology measurement system composed of multiple vantage points, and use this system to perform topological measurement studies on the Internet of mainland China; by developing the measurement system and probing the real-world Internet, we will verify, evaluate and improve our proposed key techniques, and the obtained topological data will be publish for supporting further researches and applications based on Internet topologies.

互联网的拓扑结构，对准确理解和模拟互联网，优化互联网上大规模服务应用，有重要的意义。然而，现有的拓扑测量方法存在测量不全面和冗余测量的缺陷，并缺乏准确映射入网点拓扑结构的有效方法。本项目拟重点研究互联网拓扑测量和映射中存在的关键技术问题，具体包括：（1）设计基于BGP路由嵌套的测量目标获取算法和多测量源协作的测量任务调度分配机制，实现高覆盖、低冗余的拓扑测量；（2）提出基于地理定位的入网点拓扑结构映射方法，通过定位节点的地理位置映射获取入网点拓扑结构，提高拓扑的完整性和准确性；（3）基于入网点拓扑分析互联网的结构和路由行为特征，建立路由模型，并提出入网点路径预测算法，为管理和优化大规模互联网内容服务提供理论和技术基础；（4）实现一个多点拓扑测量平台，对真实的中国大陆互联网开展周期性测量，检验、改进关键技术，获取并发布拓扑测量结果数据，为准确理解、模拟和有效管理互联网提供技术和数据支持。

全面、高效地获取互联网的拓扑结构一直是互联网测量的难点。现有互联网拓扑测量方法存在测量不全面、冗余测量、测量结果难以准确映射为入网点拓扑结构的问题。本项目重点研究了互联网拓扑测量和映射中的关键技术，具体包括：（1）提出了基于BGP路由嵌套的测量目标获取算法和多测量源协作的测量任务调度分配算法，实现了高覆盖、低冗余的拓扑测量；（2）提出了基于地理定位的入网点拓扑结构映射方法，通过定位节点的地理位置映射获取入网点拓扑结构，提高拓扑的完整性和准确性；（3）基于入网点拓扑分析互联网的结构和路由行为特征，建立了路由模型，并提出了入网点路径预测算法，为管理和优化大规模互联网内容服务提供理论和技术基础；（4）实现了一个多点拓扑测量平台，对真实的中国大陆互联网开展周期性测量，检验、改进关键技术，获取并发布拓扑测量结果数据，为准确理解、模拟和有效管理互联网提供技术和数据支持。本项目的研究成果，对准确理解和模拟互联网、优化互联网上大规模服务应用，有重要的意义。