Machine Learning Model for Detecting Greedy Node in Lorawan: A Framework

Abstract

The energy prediction and monitoring in a smart city is very crucial as many devices are involved in it to making it smart and sustainable. Long-range connection, low cost, and power-restricted user equipment are made possible by LPWAN networks across a wide area. For these networks to be used in applications for smart cities, its limited battery life is a challenge. This work primarily explores to foresee the energy consumption of user nodes that would reach the receiver end and provides web application as user interface. Employing various machine learning algorithms such as regression, and tree-based methods, enable us to capture complex relationships and patterns in energy data. This allows to make precise predictions about both short- and long-term energy needs. An optimal model is selected that best fits the criteria. In this study, the Gradient Boost provides a minimum RMSE value and better R-squared value of over 0.96 in comparison to other machine learning (ML) models. Through this prediction model, the smart city applications are benefited greatly as they support the upkeep of a centralized system to track the energy usage of user nodes.