Measuring the resilience of mobile ad hoc networks with human walk patterns

Abstract

MANET (mobile ad hoc network) technology has become increasingly attractive for real-world applications in the past decade. Dynamic and intermittent connectivity caused by node mobility poses a huge challenge to the operation of MANETs that require end-to-end paths for communication. The attacks against critical nodes could result in a more degraded network service. In this paper, we evaluate the network resilience of real-world humans' walking traces under different malicious attacks. We propose a new flexible attack strategy by selecting different centrality metrics to measure node significance according to network topological properties. We employ a resilience quantification approach to evaluate the node pair communication ability spanning a range of network operational states. Resilience of topological robustness is evaluated for different combinations of network parameters, and resilience of application layer service using different routing protocols are compared given a range of states of topological flow robustness. Our results show that flexible attacks impact overall network resilience more than attacks based on any single centrality metric with varying network connectivities.