Projection Pursuit Based Wormhole Detection in Ad Hoc Network

摘要

Ad hoc networks have found increasingly wide applications in many fields nowadays. At the meantime, various attacks against Ad hoc have emerged, such as wormhole attack, denial of service (DoS), black holes attack and gray holes attack, among which wormhole attack is most difficult to detect and is most destructive to the network. At present, there are limitations in various worm hole detection mechanisms, some require that the data collected should conform to specific probability distribution, and some even require the introduction of encryption and authentication mechanism or additional equipment and are poor in flexibility. Projection pursuit (PP) based wormhole detection mechanism does not depend on specific data distribution and can directly mine data dependency for wormhole evaluation on network nodes. PP adopts genetic algorithm to optimize projection direction and the projection results obtained can directly be used in wormhole evaluation. In this paper, we first simulate AODV routing protocol based Ad hoc network by using NS2.After data collection and through comparing a one dimensional projection and two dimensional projections, we find that one-dimensional project has little data loss and high accuracy in wormhole detection. Projection pursuit based wormhole detection gives us new insights into wormhole detection and is a wormhole detection completely based on statistics.