Policy-Preferred Paths in AS-level Internet Topology Graphs

• Main Author: Mehmet Tozal
• Format: Article
• Language: English
• Published: Georgia Southern University 2018-03-01
• Series: Theory and Applications of Graphs
• Subjects: Internet Topology Graphs; Algorithms; Policy Consistent Paths
• Online Access: https://digitalcommons.georgiasouthern.edu/tag/vol5/iss1/3

Using Autonomous System (AS) level Internet topology maps to determine accurate AS-level paths is essential for network diagnostics, performance optimization, security enforcement, business policy management and topology-aware application development. One significant drawback that we have observed in many studies is simplifying the AS-level topology map of the Internet to an undirected graph, and then using the hop distance as a means to find the shortest paths between the ASes. A less significant drawback is restricting the shortest paths to only valley-free paths. Both approaches usually inflate the number of paths between ASes; introduce erroneous paths that do not conform to economic policies; and generate symmetric paths, which in reality is not a rule. As a result, the derived conclusions might be greatly misleading. In this study we introduce a single-destination, policy-preferred path enumeration algorithm which discovers policy consistent paths from all ASes to a destination AS in an edge-labeled, finite, directed graph representing the Internet topology. Considering that our algorithm's run time complexity is the same as Dijkstra's shortest paths algorithm, we believe that the proposed algorithm will notably enhance the future works that leverage AS-to-AS paths in Internet topology graphs.