Machine Learning based Improved Gaussian Mixture Model for IoT Real-Time : Data Analysis
Introduction: The article is the product of the research “Due to the increase in popularity of Internet of Things (IoT), a huge amount of sensor data is being generated from various smart city applications”, developed at Pondicherry University in the year 2019.
Problem:To acquire and analyze the huge amount of sensor-generated data effectively is a significant problem when processing the data.
Objective: To propose a novel framework for IoT sensor data analysis using machine learning based improved Gaussian Mixture Model (GMM) by acquired real-time data.
Methodology:In this paper, the clustering based GMM models are used to find the density patterns on a daily or weekly basis for user requirements. The ThingSpeak cloud platform used for performing analysis and visualizations.
Results:An analysis has been performed on the proposed mechanism implemented on real-time traffic data with Accuracy, Precision, Recall, and F-Score as measures.
Conclusions:The results indicate that the proposed mechanism is efficient when compared with the state-of-the-art schemes.
Originality:Applying GMM and ThingSpeak Cloud platform to perform analysis on IoT real-time data is the first approach to find traffic density patterns on busy roads.
Restrictions:There is a need to develop the application for mobile users to find the optimal traffic routes based on density patterns. The authors could not concentrate on the security aspect for finding density patterns.