Comprehensive study of the response of inexpensive low energy wireless sensors for traffic noise monitoring

Shadman Sadik Marouf*, Margaret Carol Bell, Paul Stuart Goodman, James Neasham, Jeffery Neasham, Anil Kumar Namdeo

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

8 Citations (Scopus)

Abstract

Precision noise monitoring can prove to be prohibitively expensive and labour intensive. Low cost, wireless, environmental sensor systems, referred to as motes, are being developed as alternatives. Whilst considered of lower accuracy than precision sound level meters, they can be deployed for much longer periods and monitor continuously at one minute resolution. This paper reports the performance of two commercially available motes, the eMote developed to monitor traffic and the nMote to monitor airport noise, in both controlled laboratory conditions, and deployment in the field. Under controlled conditions the performance was assessed by co-location with precision sound level meters (Brüel and Kjær Type, 2260) and exposed to coloured (‘white’ and ‘pink’) noise at different levels. Both units exhibited strong linear relationships (R2 > 0.98) for LAeq when compared to the precision instruments. The performance of three individual eMotes, in terms of similarity and variability, were broadly consistent across the tested units, though levels were systematically overestimated by approximately ±3 dBA when uncorrected. The eMote units were found to exhibit a lower detection limit of ~55 dBA, potentially limiting their application to (a) urban daytime conditions when levels generally exceed that threshold and (b) at other times and locations as screening tools for excessive noise levels, to justify more expensive precision monitoring confirming findings of Marouf et al. (2018). Further research here demonstrates that the systematic error could be corrected by an algorithm resulting in a random error of better than ±1.5dBA. Both nMote units exhibited a lower detection limit (~40dBA). One nMote was electronically calibrated at the manufacturing stage and found to slightly underestimate at lower and overestimate at higher levels. Some improvement was made by developing an algorithm giving overall measurement accuracy better than ±2dBA. The other nMote not calibrated at the manufacturing stage underestimated values at low levels. This could be corrected resulting in a random error of better than ±2.5dBA. Similar results for both Mote types were observed in the field but with more variability due to the data synchronisation, local reflection and refraction etc. and the continually varying noise characteristics of traffic. Whilst the nMote was designed for airport noise, it is considered along with the eMote suitable for monitoring traffic noise in urban areas, and demonstrates much potential to be incorporated into existing Intelligent Transport System technologies to measure, manage and control population exposure to traffic noise (e.g. as a ‘noise camera’ analogous to the existing speed camera). Given that deriving correction algorithms for motes in controlled test environments is impractical, the knowledge gained from these studies has led to modifications to the design and with electronic calibration as part of the manufacturing process measurement errors of better than ±2dBA across the range of sensitivity can be achieved. With advancements in technologies and if power consumption was not an issue, then nowadays a device could quite easily be designed with a dynamic range and linearity quite close to the precision instruments for little extra cost.

Original languageEnglish
Article number107451
JournalApplied Acoustics
Volume169
Early online date20 Jun 2020
DOIs
Publication statusPublished - 1 Dec 2020
Externally publishedYes

Keywords

  • Noise pollution
  • Pervasive monitoring
  • Road traffic noise
  • Wireless sensors (motes)

Fingerprint

Dive into the research topics of 'Comprehensive study of the response of inexpensive low energy wireless sensors for traffic noise monitoring'. Together they form a unique fingerprint.

Cite this