A Machine Learning Solution Based on a Cloud Computing Platform Dedicated to IoT-IA Behavioural Analysis
Nour El-Houda Benalia, Kheira Lakhdari, Souraya Hamida, Rabah Sadoun · 2025
The Internet of Things (IoT) seeks to interconnect numerous devices and tools with the Internet, enabling greater control over the physical world while making it smarter and more responsive to human needs. A key challenge of IoT technology is the collection and analyzing of large amounts of data to develop AI-driven systems capable of making autonomous decisions. This paper presents a software architecture that supports an AI analysis platform designed for IoT applications. The platform leverages Big Data and Cloud-based architecture to enhance performance in managing large datasets and complex computations. It enables high-performance applications that address socioeconomic issues such as traffic management, water consumption, and energy optimization. The proposed solution will be tested in a real-world scenario, focusing on the behavioral analysis of 4-wheel vehicle drivers to develop a driving classification model that assesses whether the behavior is indicative of safe driving. Such a system has potential benefits for public authorities and insurance companies, offering a means to improve safety and reduce costs. In our study, we classify driving behaviour as dangerous or not according to speed and engine RPM. To achieve this, we implement a scoring system that assigns a risk score to each type of vehicle, quantifying its level of danger. Experimental results demonstrate that feedback control efficiently fulfils diverse user resource demands, enhances user satisfaction, and maximizes system resource utilization.