Drive-by Multi-frequency Large-region Noise-source Mapping via Tomographic Imaging

Recent studies indicate that noise pollution has hazardous impacts on human health and behavior, particularly becoming a serious problem in densely-populated cities. In order to effectively deal with the environmental noise problem on a city-wide scale, there is a great need for a new low-cost environmental noise-monitoring system, which will replace conventional sound-pressure-level (SPL) meter-based techniques that require overly expensive microphone deployment at a dense grid of locations while providing only long-term average noise levels. In this paper, we propose an advanced cost-effective large-region environmental noise mapping scheme that uses a microphone array mounted on a vehicle driving around the city to make detailed acoustic noise maps without the need for point-by-point measurements. We demonstrate that the multi-frequency SPLs and locations of sparsely-distributed localized noise-sources in the neighborhood traversed by the vehicle are jointly estimated via tomography, in which the far-field delay-and-sum beamforming output power values computed at multiple locations and different times as the vehicle drives past are used for tomographic imaging. The preliminary drive-by noise-source mapping experiments with two circular arrays rooftop-mounted on an electric vehicle shows the promise of this new proposed technique for cost-effective city-wide large-scale environmental noise monitoring.