Analysis of Cascading Failure Based on Wireless Sensor Networks

Research relating to the invulnerability of Wireless Sensor Networks (WSNs) has made gratifying progress. However, most of them concentrate on the statistic features of networks, and ignore the cascading failure of network caused by dynamic load changes. In this study, considering the realistic characteristics of WSNs, random network, scale-free network, WS small-network and NW small-network models have been built and the invulnerability performance and cascading process of these network models under random attack are researched respectively. The study presents that the increase of the coefficient of tolerance-T is beneficial to improving the invulnerability of all networks, especially NW small-world network model. By evaluating the distribution of causes (i.e., traffic overload, invalid connectivity) to node failures, it was found that invalid connectivity is major reason for failure nodes. Besides that, scale-free network shows more steady performance than other networks in terms of error-tolerance.