Machine Learning application to IoT and IIoT security and reliability

The Internet of Things (IoT) is a network of interconnected devices and objects spanning various domains, including smart homes and industrial en- vironments. Machine learning (ML) technologies often enhance this ecosys- tem, facilitating transformative potential through automation, real-time data processing, and efficient resource management. By optimizing processes, ML algorithms provide solutions that address data management, predictive an- alytics, anomaly detection, and cybersecurity. In industrial contexts, ML contributes to predictive maintenance and operational efficiency, while in networking, it enhances traffic analysis and threat detection. The applica- tion of ML enables IoT systems to adapt to evolving challenges, thereby improving resilience. However, the implementation of these technologies presents challenges. Hardware and software limitations frequently vary by application field, and the successful adoption of standard data models depends on widespread ac- ceptance. Moreover, the proposed ML approaches are typically tested offline, which may not fully represent the complexities of real-world scenarios. This thesis investigates effective machine learning techniques within industrial and networking contexts to improve data management and enhance security by analyzing large datasets and adapting to changing patterns. Emphasis is placed on the necessity of data model standards for efficient data flow man- agement, interoperability, and scalability. Additionally, the focus is directed toward deploying machine learning-based solutions for malware detection, highlighting the importance of addressing data constraints.