LOST: A Location Estimator Scheme for PM2.5 Pollution Sources in Sparse Sensors Network

• Main Authors: Ghaffar, Faisal, Kibirige, George William, Shen, Chih-Ya, Chen, Meng Chang
• Format: Conference Proceeding
• Language: English
• Subjects: Computational modeling; Dispersion; Estimation; Internet of things; Monitoring; PM2.5; Pollution; Pollution dispersion; pollution source; Position measurement; sensor network; Sensors
• ISSN: 2576-6813
• DOI: 10.1109/GLOBECOM42002.2020.9348245

To protect people from hazardous pollution exposure, mainly Particulate Matter of diameter 2.5 microns or less (PM2.5), countries continuously monitors the air quality via air quality monitors and internet of things based sensors. However, these monitors and sensors are deployed at sparse locations that pose the challenge of estimating the location of the PM2.5 pollution sources. To cope with this challenge, we propose a location estimator (LOST) scheme, which locates PM2.5 pollution sources in a sparsely deployed sensors environment. LOST efficiently models the spatio-temporal dispersion of PM2.5 of each pollution source, which enables LOST to compute the strength of the surrounding sensors based on the wind direction and sensors-source proximity. LOST utilizes a robust approach to backtrack the PM2.5 pollution sources via the weighted spatiotemporal strength of sensors' PM2.5 concentration. Compared to the state-of-the-art estimation schemes for source location, LOST shows improved performance. Our experiments show that, on average, LOST, attains higher closeness and lower failure ratio by more than 16.2% and 6.5%, respectively.