Large-region dynamic tomographic mapping of environmental noise sources with microphone arrays

Environmental noise has become a major problem in large urban cities, raising concerns on mental and physical human health. However, current noise-monitoring techniques may not capture the intermittent annoying variations in the acoustic field due to very loud nonstationary noise sources such as banging from construction sites or traffic noise from moving vehicles. We propose a dynamic tomographic imaging approach for large-region acoustic mapping of nonstationary noise sources using the far-field delay-and-sum beamforming output power of prepositioned microphone arrays as tomographic measurements. We formulate the dynamic tomographic image formation under the linear state-space model, in which we recursively estimate the unknown acoustic field via a Kalman filter given the measurements. Preliminary dynamic acoustic mapping experiments with two fixed eight-channel circular microphone arrays demonstrate that with this new tomographic approach, it becomes possible to effectively characterize individual nonstationary acoustic noise sources on 2-D consecutive frame-by-frame noise-maps, offering an alternative advanced environmental noise-monitoring solution.