An IoT-Based Environmental Monitoring System Using Adaptive Filtering Techniques

In this study, we developed a proof-of-concept IoT-based Environmental Monitoring System (EMS) using adaptive filtering methods to improve the quality of the collected signals. The EMS contains inexpensive air quality, temperature, humidity, light level sensors interfaced to an ESP32 microcontroller along with cloud-based analytics, and was tested over 14 consecutive days in a variety of locations across five different urban environments. The primary innovation is a new algorithm called the HAF which demonstrated improvements in signal quality over every measure when compared to a single, unfiltered measurement as well as compared with independent signal processing methods such as Kalman (which showed 38%-45% improvements in MSE and 8-12 dB improvements in SNR) and LMS filters (which showed 25%-30% improvements in MSE and 6-8 dB improvements in SNR). In addition, HAF's mode-switching feature proved particularly beneficial at high variability sites (up to 65% improvement in MAE at the industrial site). The new EHI developed using HAF processed multiple sensor data streams into one composite score which was validated against the regulatory AQI (r = 0.89, p < 0.001). This represents the first time a composite environmental index based on adaptive-filtered signals from IoTs has been developed within the literature. Analysis of the diurnal and site-stratified datasets show that HAF mode-switching behaviors follow clearly intuitive environmental patterns and confirm that the algorithm is effectively detecting and adapting to signal regime changes across a wide variety of urban settings, as intended. All filtering configurations were found to have negligible impacts on the battery life of the system (i.e., maximum 5-day loss from the normal estimated 37–42-day cycle) and demonstrated acceptable real-time latency when processing data (i.e., maximum 3.9 ms), suggesting they can be practically deployed on resources-limited embedded systems. Overall, the results from this study suggest that there is sufficient evidence to support the claim that utilizing low-cost IoT sensing technologies in conjunction with adapted signal processing, principles of composite index health monitoring, and consistent accuracy allows for reasonably comparable accuracy and intelligibility when developing large-scale environmental monitoring systems; thus, allowing for the development of dense metropolitan-scale monitoring networks to address 21st century air quality and climate issues.

[1]  L. Zhang and J. Zhang, “IoT-enabled air quality monitoring and filtering for urban areas,” Sensors, vol. 19, no. 7, p. 1662, 2019. [2]  A. Kumar, P. Singh, and R. Kumar, “Noise characterization and compensation in low-cost IoT environmental sensors,” IEEE Sensors Journal, vol. 21, no. 14, pp. 15820–15831, 2021. [3]  H. Li, W. Chen, and X. Zhou, “Adaptive signal processing for IoT sensor networks: A survey,” IEEE Communications Surveys & Tutorials, vol. 23, no. 2, pp. 1025–1058, 2021. [4]  B. Widrow and S. D. Stearns, Adaptive Signal Processing. Englewood Cliffs, NJ: Prentice-Hall, 1985. [5]  I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “Wireless sensor networks: A survey,” Computer Networks, vol. 38, no. 4, pp. 393–422, 2002. [6]  U. Raza, P. Kulkarni, and M. Sooriyabandara, “Low power wide area networks: An overview,” IEEE Communications Surveys & Tutorials, vol. 19, no. 2, pp. 855–873, 2017. [7] A. McEwen and H. Cassimally, Designing the Internet of Things. Chichester, UK: Wiley, 2014. [8] S. M. Kay, Fundamentals of Statistical Signal Processing: Estimation Theory. Englewood Cliffs, NJ: Prentice Hall, 1993. [9] M. Snyder et al., “The changing paradigm of air pollution monitoring,” Environmental Science & Technology, vol. 47, no. 20, pp. 11369–11377, 2013. [10] E. Spinelle et al., “Field calibration of a cluster of low-cost commercially available sensors for air quality monitoring,” Sensors and Actuators B: Chemical, vol. 215, pp. 249–257, 2015.