Local governments strive to make city centers an enjoyable place to live, for example, by organizing events and providing numerous facilities for transport and cultural, recreational and sports activities. Both citizens and city officials wish to keep youth centers, cultural venues, playgrounds, and parks near the city center as it makes them easily accessible to the general public. However, this often creates new challenges, especially in terms of noise nuisance. Sound pollution has been recognized as a common problem in urban environments. Moreover, according to Basner et al. (2014) and Hammer et al. (2013), it has increasingly been shown to affect public health and the overall quality of life. Recent reports of the European Environment Agency on noise are limited and mainly based on simulated noise maps obtained from large-scale modelling. As updates of these noise maps can take up to six or more years, these maps are fairly static and provide only an indication of the long-term average noise level.
There are two approaches to map noise pollution in urban areas. The first approach relies on volunteers that use their mobile phones for real-time Sound Pressure Level (SPL) measurements. The quality and reliability of these crowdsourced measurements are questionable as no proper calibration process is in place. The second approach consists of Wireless Acoustic Sensor Networks (WASNs), offering large-scale coverage while producing continuous calibrated acoustic measurements. Current systems have limitations in terms of deployability and high-power consumption.
This IOASIS-project is part of the latter, a WASN approach. Our aim is to provide up-to-date insights into sound pollution and instantaneously detect acoustic events, while keeping the power consumption low and maintaining the privacy of the end-users. The monitoring system can act as an unbiased source of information in addition to human complaints and can notify the responsible authorities in case of continuous or suspicious sound events. The main goal is to enrich urban areas with oases of tranquility.
The low-power, smart sensing network consists of ultra-low-power, low-cost and easily deployable sensor nodes and a central gateway. The nodes are spread out over an area of interest and evaluate the presence of noise nuisance based on three signal characteristics: the sound intensity (SPL), the repetitiveness, and the frequency spectrum. The proposed implementation differs from previous solutions through discontinuous audio sampling implemented by a wake-on-sound MEMS microphone. By limiting wake-up events and implementing local processing before data transmission, the power requirements of the nodes are kept to a minimum. This allows for battery powered operation, facilitating easy-deployability in the field. After detection and local processing of the noise nuisance, the extracted audio features are transferred to a central entity. In this way, data of multiple sensing nodes can be combined to identify the sound source and its location. The Long Range Wireless Area Network (LoRaWAN) communication technology was chosen to take on this role.
A general block diagram of the sensing node is depicted in the additional pictures. A MEMS microphone followed by an amplifier-filter circuit is connected to an EFM32 Happy Gecko microcontroller board of Silicon Labs. Communication with the LoRaWAN gateway is established through a RN2483 LoRa transceiver. As a microphone, the Vesper VM1010 is selected.
The prototype of the mobile node is combined of off-the shelf chipsets and in-house designed printed circuit boards. This guarantees the low-power consumption of the audio front end. For large-scale productions, all entities can be integrated in a lightweight single board, reducing the cost and increasing the deployability even more.
The central entity is comprised of a commercially available LoRa-gateway for data reception linked with any type of central processing unit. The IOASIS project has a direct impact on the following urbanization problems:
Public health. Noise pollution has a direct impact on public health. By deploying these networks, local authorities have an unbiased source and can act immediately if necessary. Thanks to its deployability and low maintenance, more profound research can be performed on the impact of noise on the public health.
Energy consumption. The low-power sensing nodes have an estimated power consumption of 183.646 mJ for a single transfer (946 µJ for the audio front end and 182.70 mJ for the RF data transfer). Current WASN networks are connected to the power grid and power usage is not one of the main concerns. Our vision is better to prevent than to cure, in other words, it is better to cut the power than to use (green) power.
Safety. The responsible authorities can be warned in case of suspicious sound events. Combining the data of several sensing nodes can give an indication of the sound event, reducing the area of interest for potential search and rescue operations.
Privacy. Due to local processing of the audio data, all of the above-mentioned solutions maintain the privacy of the audio subjects. No recorded audio can be intercepted from the data communication.