| Abstract: | Intrusion detection is prominently important for civil and military applications in wireless sensor networks (WSNs). To date, related works address the problem by assuming a straight‐line intrusion path and a Boolean sensing model. However, a straight‐line intrusion path is often not the case in reality, and the Boolean sensing model cannot resemble a real‐world sensor precisely. Results based on these assumptions are therefore not applicable with desirable accuracy in practice. In view of this, we propose a novel sine‐curve mobility model that can simulate different intrusion paths by adjusting its features (amplitude, frequency, and phase) and can be integrated into the random WSN model for intrusion detection analysis. It can also be applied to different sensor models and makes influencing factors tractable. With the model, we examine the effects of different intrusion paths on the intrusion detection probability in a random WSN, considering both Boolean and realistic Elfes sensing models. Further, we investigate the interplays between network settings and intruder's mobility patterns and identify the benefits and side effects of the model theoretically and experimentally. Simulation outcomes are shown to match well with the theoretical results, validating the modeling, analysis, and conclusions. Copyright © 2011 John Wiley & Sons, Ltd. |
