CSTEP’s work on assessing the performance of low-cost sensors manufactured in India is being used by Google for Project Air View. As part of the research, CSTEP tested the performance of 50 air sensors from six Indian manufacturers by comparing their data against that of reference-grade monitors. The tests were conducted over a period of 6 weeks at the India Sensor Evaluation and Training Center (Indi SET) set up at CSTEP. The study looked beyond PM2.5 measurements to trace gases as well. Using the data from this study, we developed our own calibration model, employing machine learning algorithms to improve the accuracy of the sensor data. We also established a multi-pollutant low-cost sensor network across the city of Bengaluru, informed by satellite estimates of surface air pollution, land-use attributes, and identification of various point-of-interest areas. The data from the sensors is shared with Google every 15 minutes, which is then used to provide almost real-time information on pollution levels in the city.
