Edge-Based Data Sensing and Processing Platform for Urban Noise Classification

The increasing population in urban areas contributes to noise pollution. So, cities look to locate areas with high noise levels to regulate them and improve urban well-being. However, dispatching personnel for noise data collection is time-consuming and expensive. Therefore, we propose a low-cost Internet of Things (IoT)-based urban noise classification platform to address noise collection and processing challenges in urban environments. We designed a prototype sound-sensing setup consisting of an STM32 NUCLEO-64 board with an attached X-NUCLEO-CCA02M1 expansion board as the digital micro-electrical-mechanical system microphone connected to a Raspberry Pi for collecting and classifying urban sound. Then, it sends the results via WiFi to the cloud server for noise analysis. We examined our design's feasibility with experiments evaluating its classification accuracy and power consumption. We further examined its latency when processing is at the edge or cloud. The experimental results suggest our platform's potential in noise analysis for urban environments.