From Malicious Eyes: A Method for Concise Representation of Ad-Hoc Networks and Efficient Attack Survivability Analysis.

摘要

Ad-hoc networks are undergoing widespread use because they are able to provide capabilities that are infeasible with traditional infrastructure networks such as adaptive topologies. With these capabilities, however, security risks and resource limitations are introduced. Much work has focused on the development of security strengthening mechanisms such as secure routing protocols, traffic encryption, distributed intrusion detection, and others. Many times, these methods are not feasible due to limitations of available processing and power. In addition, it has long been known that security can never be guaranteed. Alongside security analysis, survivability analysis focuses on the ability of network entities to function even during attacks. While previous methods attempt to measure system tolerance as a whole, for example, average throughput over several simulations, still missing are methods that are able to predict low-level attack impacts. Analysis of these low-level impacts enables analysts to tune and redesign networks to optimize survivability. In this report, a dataset is collected and formatted into a novel network representation. This representation is then used to build a classifier that accurately predicts link loss due to spoofing and data forwarding attacks.