Evaluating and Improving Regional Network Robustness from an AS TOPO Perspective

• Published in: IEEE/IFIP Network Operations and Management Symposium pp. 1 - 7
• Main Authors: Liu, Yujia, An, Changqing, Yu, Tao, Zheng, Zhiyan, Pei, Zidong, Wang, Jilong, Charnsripinyo, Chalermpol
• Format: Conference Proceeding
• Language: English
• Published: IEEE 08.05.2023
• Subjects: Autonomous System (AS); network measurement; network resilience
• ISSN: 2374-9709
• DOI: 10.1109/NOMS56928.2023.10154451

Currently, regional networks are subject to various security attacks and threats, which can cause the network to fail. This paper borrows the quantitative ranking idea from the fields of statistics and proposes a ranking method for evaluating regional resilience. Large-scale simulated failure events based on probabilistic sampling is performed, and a significance tester that measures the impact of events from the overall level and variance aspect is also implemented. To improve a region's robustness, this paper proposes a greedy algorithm to optimize the resilience of regions by adding key links among AS. This paper selects the AS topology of 50 countries/regions for research and ranking, evaluating the topology robustness from connectivity, user, and domain influence perspectives, clustering the results and get typical region types, and adding optimal links to improve the network resilience. Experimental results illustrate that the resilience of regional networks can be greatly improved by establishing a few new connections, which demonstrates the effectiveness of the optimization method.