Design and Implementation of an IoT for Flood Prediction

Abstract

  This thesis proposes a comprehensive system designed to address the challenges associated with efficient data collection from diverse Internet of Things (IoT) sensors for flood monitoring by providing a scalable and robust Internet of Things (IoT) platform.

The IoT environment for flood monitoring and prediction system was designed to promote sustainability and autonomy by preferring sensors with minimal footprints and compatibility with solar panels. The system architecture incorporates the Message Queue Telemetry Transport (MQTT) protocol, enabling seamless integration with snow sensors. It also leverages a 4G network for seamless data transmission. We propose designing and implementing a centralized system called HYDROSIGHT, allowing for real-time collection, monitoring, and visualization based on sensor data for flooding. HYDROSIGHT platform also includes a log monitoring feature for effective debugging and troubleshooting.

In order to validate the practical applicability of the proposed system, HYDROSIGHT were developed at two municipalities of Quebec, namely Terrebonne and Lac-Superior. In addition to this, testbed at Ericsson facility in Montreal was developed to test the 5G capabilities of the system. The real-world deployment in these locations allowed us to evaluate the performance and effectiveness of the HYDROSIGHT system in a practical flood monitoring scenario.

In addition to developing the IoT system, a preliminary experiment is conducted on forecasting water level heights. The online machine-learning approach was used for this experiment. By utilizing real-time data from the HYDROSIGHT system, we trained and tuned an online machine learning model for accurate water level forecasting.

Furthermore, a comparative analysis is performed between traditional batch machine learning and online machine learning approach for water level forecasting with time series data from HYDROSIGHT system deployed at Lac-Superior in Quebec. This analysis provides insights into each methodology’s advantages and limitations in water level forecasting.