Cross-Correlation Based Vehicle Feature Extraction by Magnetic Wireless Sensor Networks

—Due to the limitation of sensor size and cost, in order to implement vehicle detection by magnetic wireless sensor network, it attaches importance to the two important problems, vehicle detection and feature extraction, with facing the challenge of low signal-to-noise ratio and limited resources. In this paper, we propose a mechanism to accomplish vehicle monitoring, namely CBNP. In this mechanism, we propose a cross-correlation based vehicle detection algorithm to accurately detect vehicle presence. To accurately extract response feature, we propose and prove a theorem which depicts the relation between feature point and its cross-correlation result. Based on the theorem, we proposed a Cross-Correlation Based Feature Extraction (CBFE) algorithm to accurately locate vehicle feature points under relative strong noise. Simulation indicates that CBNP outperforms traditional methods on both vehicle detection rate and feature extraction accuracy. In meaningful conditions, the detection accuracy is over 90% and the deviation rate of feature extraction is within 10%.