Time Domain Energy Detector approach for Detection of Moving Ground Vehicle

Abstract

This paper introduces a time domain energy detector (ED) approach for vehicle detection using a single microphone. The detection of a deterministic, as well as the stochastic signal in the presence of white Gaussian noise (WGN), is a very challenging task. The ground vehicle generates a similar kind of signal with unknown information. Therefore, there is need of robust vehicle detection algorithm which must detect a vehicle with low false alarm. We proposed an algorithm based on energy detector and estimation of threshold value depending on noise s variance for moving vehicle detection using acoustics data. The energy distribution of the signal follows the chi-square distribution. These decision statistics reveal the vehicle presence information from the signal and detection probability of a true event. The algorithm is tested on the experimental data acquired from the field trials of the moving vehicle. The receiver operating characteristics are computed through Monte Carlo simulations for validation of the proposed algorithm. The results show that the algorithm is capable to detect the moving vehicle with low false alarm even at low signal to noise ratio.