Crowd sensing based urban noise map and temporal-spatial feature analysis

The noise map can intuitively present the urban noise level by encoding levels with colors. To get noise map efficiently and economically, a crowd sensing based noise-mapping system is developed. First, a noise level measurement is performed on mobile phones. A server aggregates all the measurements to get the complete noise pollution information so that the noise map can be viewed on mobile phones. The temporal-spatial features of the noise distribution in cities are analyzed and discussed. To solve the problem that noise samples are random and incomplete, a novel data sampling approach based on compressive sensing is proposed. Our approach does not need inter-sensor communication in the sense that every node samples the field randomly and independently, and the noise data ensemble can be reconstructed with fairly high quality when all compressed data are used simultaneously. Experimental results indicate that the system can reduce the communication overhead with acceptable accuracy.