Detection of Various Vehicles Using Wireless Seismic Sensor Network

The Detection of vehicles, whether wheeled or tracked can be achieved using seismic sensors as all vehicles when they move create vibrations which travel as waves through the earth and based on the type of vibrations generated by the movement of vehicles; classification of the vehicles can be carried out. This can also be achieved by using acoustic sensors but this approach inherits the problem of Doppler shift which is also associated with detection of signals through Radars. This paper is devoted to seismic detection of vehicles by using unpowered geophones sensors. The event triggering is captured in real time in different terrain scenarios like loose earth or concrete approaches. The geophones are connected through wires with the event detection and data analyzing system. However, the complexity of the seismic waves and its dependency on the underlying geological characteristics throws the challenges in the vehicle detection and recognition and extraction of robust feature vector which can explicitly correspond to a specific type of vehicle or target. This paper has emerged from the series of field experiments carried out and performance evaluation conducted to derive the optimal relationship of the different parameters. It proposes a new feature extraction algorithm- spectral statistics and wavelet coefficients characterization (SSWCC). SSWCC extracts a feature vector from both the frequency and the time frequency domain analysis of the seismic signals, including the spectrum, the power spectral density (PSD) and the wavelet coefficients. The SSWCC algorithm is designed for real-time applications, and a series of experiments shown its robustness and effectiveness.