Robust Fusion of Multiple Microphone and Geophone Arrays in a Ground Sensor Network
Abstract
A sensor network with arrays of microphones and geophones has been deployed along a track in the country. Together with a beam forming technique, the sensor arrays produce bearing estimates to passing vehicles. The bearing estimates are sent to, and fused by, a Kalman tracking filter. We report how we implemented this system with low-complexity distributed-type hardware and software components, and also indicate the level of performance to be expected using a small number of intelligent network nodes. We also study a large data base with acoustic recordings collected in realistic environments with the purpose to classify vehicles traversing the network. Support vector machine classifiers, taking as input spectral features picked up from the emitted audio, are fused in time and space to identify the vehicle model.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 2006
- Accession Number
- ADA478760
Entities
People
- A. Lauberts
- D. Lindgren
- H. Habberstad
- M.-L. Holmberg
Organizations
- Swedish Defence Research Agency