The Design And Implementation Of A Real-time Fine-grained Noise Sensing System Based On Participatory Sensing

Excessive exposure to noise pollution may cause damage to both physical and psychological health. People desperately want to know the real-time and history noise level of the area they care about. Unfortunately, it is generally difficult for ordinary people to access the noise level information because of limited noise information sta-tions and increased burden of carrying professional noise level meters. The noise level meter can only provider the noise level near users.On the other hand, with the development of science and technology, the price of the smartphone is lower and lower, and the function is more and more powerful. The penetration rate of smartphone is becoming high. Being equipped with a high-quality microphone, a smartphone can potentially serve as a handy noise level meter. With the help of GPS and internet module, smartphone users can share the noise data near them to other users, which forms a noise sensing system.However, before turning such a noise sensing system into reality, we need to over-come several challenges. First of all, the straightforward use of sound measurements from smartphones leads to large measurement errors. Second, noise data contributed by users via smartphones may be sparse over a wide range of space and time. Finally, in order to serve plenty of users, the system needs to working properly under huge ca-pacity and processing plentiful data within some period of time. To decrease the big measurement error, we find there is a linear relationship between the measured values of smartphones and the readings of the standard noise level meter. We also discover that the noise level of a quiet indoor environment is approximately a constant value. We put forward a lightweight calibration algorithm, and experimental results show that the calibration error is as low as 3 dbA compared to professional noise level meters. According to the sparsity of noise data, we abstract the area we cares about into matrix and construct the characteristic matrix. Matrix decomposition is used to restore the missing data. In order to support large-scale users to upload data and query results, we introduced the storm flow processing framework, and the system is designed to be a reliable and scalable distributed system.