Abstract
This dissertation presents a robust air and water quality monitoring system (AWQMS). I describe
the hardware design and implementation of scalable wireless sensor nodes (WSNs) that can be
integrated with artificial intelligence (AI) to forecast future pollutant concentrations. The AWQMS
uses an array of various low-cost, commercially off-the-shelf (COTS) hardware components and
high-quality sensor actuators to measure air-water quality parameters, namely NO2, SO2, O3, pH
and DO. The developed system deploys WSNs to sense selected ambient air-quality and waterquality
sensory data in a wireless sensor network topology. These smart WSNs generate real-world
data and transmit the collected sensory data wirelessly using a GSM click module to a designed
API server. The API server acts as a database and securely stores the sensory data in a cloud-based
platform. Thereafter, a graphical user interface (GUI) is developed, to allow the end-user to
perform data analytics on the sensory data using developed time-series models from AI’s machine
learning algorithms. As a result of the designed AWQMS hardware prototypes, the Auto-Arima
and Prophet time-series models are trained and fine-tuned to forecast future air-water quality
parameters with a degree of certainty, for the next 3 hours.
Keywords: Air and Water Quality Monitoring System, WSNs, sensory data, GSM click module,
API server, GUI, AI, Machine Learning, Auto-Arima, Prophet.