Predictive Maintenance for Quality of Service in IoT Communications

Abstract

The evolution of 5G networks introduces dynamic, virtualized environments where seamless connectivity and quality of service are essential. Mobility management is crucial for optimizing resource allocation, ensuring service continuity, and enhancing predictive maintenance. Machine-type communications support diverse IoT applications requiring ultra-reliable, low-latency communication. Network slicing addresses these needs by differentiating between mission-critical and massive communication slices. This thesis explores the role of mobility prediction in improving the quality of service by forecasting user transitions between network cells, enabling proactive service management. A next-cell prediction framework using Long Short-Term Memory networks is proposed. Evaluated in vehicular networks, the model outperforms conventional classifiers, demonstrating its potential to enhance predictive maintenance and resource allocation.