ABSTRACT
Air pollution poses a significant environmental challenge with both short and long-termrepercussions for human health, especially in rapidly growing urban areas like Lagos. This research aims to address this challenge by developing and implementinganair pollution monitoring and reporting system using Wireless Sensor Networks (WSN). Thefocus is on deploying WSN nodes equipped with low-cost air pollution sensors tocollect real-time data, enabling data reporting, storage, analysis, and prediction of air pollutiontrends. By assessing risks based on concentration levels, the study seeks to contributevaluable insights into the mitigation of air pollution effects and sustainable urbandevelopment. The research utilized low-cost semiconductors and electrochemical sensors for measuringpollutants, including CO, CO2, NO2, SO2, O3, and NH3. Sensor nodes were designedwithATMEGA328P, Arduino UNO boards, and ETD-868-32T LoRa wireless modules, enabling communication within a hierarchical 3-tier WSN topology. The study was donein Ojo in Lagos State, Nigeria and covers a total area of 7.52km2 (2.51 mi2) anda total distance of 10.47km and its GPS coordinates are 6.63026, 3.34569 north, 6.2011, 3.36136 East, 6.61594, 3.33544 West, 6.0016, 3.34668 South.These nodes werestrategically placed across Lagos, connected to signal conditioner circuits, and poweredby solar panels with battery backups to ensure continuous operation. MATLABwas usedto develop database to archive collected data, comprising hourly and monthly readingsfor the year 2019 and a graphic user interface (GUI) to display the data andfor processing the data as charts and trend-lines at the base station. The study employedmachine learning models like back-propagation neural network (BPNN), nonlinear autoregressive artificial neural network (NARX), and the Adaptive Neuro-Fuzzy InferenceSystem (ANFIS) and ANFIS was used finally as it provided the best fit for air pollutiontrend prediction. The AI model was trained with data from the first half of the year. Mitigation of air pollutants using drone deployment mechanism of reduction solutiontoaffected areas for area spraying activity was designed and proposed. The research findings shows pollution data for 2019 revealed worrying levels of CO, CO2, NO2, and SO2, categorized as poor according to the air quality index. O3 exhibitedamoderate range, while NH3 was classified as good. Risks assessment indicatedthat concentration levels exceeded WHO, NAAQS, and FEPA standards, particularlydetrimental to sensitive groups. The implementation of the ANFIS neuro-fuzzymodel proved effective in predicting air pollution, offering a reliable tool for future assessments. The model demonstrated a 92% similarity index when predicting the secondhalf. Mitigation of pollutants using technological approach was designed and presentedthat can be implemented in the study area of interest. The study underscores the urgent needfor mitigation strategies to counteract the adverse health effects associatedwithhazardous gases surpassing threshold levels. This research contributes to informeddecision-making for sustainable urban development and public health management byoffering a comprehensive understanding of air quality dynamics. It underscores thepivotal role of monitoring systems based on Wireless Sensor Networks (WSN) intackling current environmental challenges.
