Detecting Noise-Related Public Safety and Health Risks using Smart Sensors

A growing number of municipalities in America are turning to IoT devices to detect and identify risk events based on sound patterns. Is the technology ready?

Sensor technologies, including hardware components and AI software, are getting more cost-efficient and smarter to enable real-time classification and location for different sources of sounds (e.g., gunshots, car racing, fireworks, construction). This is important because public safety has limited resources to respond to citizen complaints or sensor alerts … and they need to confidently direct their resources to addressing those risks that are most severe. One of the biggest technical challenges is to identify the source and type of the noise apart from similar noises (noise “fingerprint”) … at the “edge”, and providing metadata and alerts over a wireless network to reduce the latency of the alert and to reduct the payload and costs for connectivity. Traditionally, sensors have had to send audio files back to the cloud in order to distinguish types of noise. We are now seeing some of the leading vendors in this market use machine-learning and edge computing to solve this challenge.

Sound sensors have been used by private industries and public agencies for the past decade to monitor industrial noise pollution, such as construction, and for gunshot detection. These sensors measure decibel levels and provide users with alerts when decibel thresholds are crossed. Cities in the U.S. are now becoming more interested in detecting vehicle loud noises, including identifying vehicles that are illegally racing, those that have modified their vehicle equipment to amplify vehicle noises (eg. muffler not working/missing), those playing loud music, and more. In many cities, there are established ordinances against extreme vehicle noises, but there are few tools for agencies to detect for violations.

SenthiSYS has partnered with leading sensor manufacturers with deep technical expertise in detecting, sourcing, and alerting for a range of noise violations.