ARTIFICIAL INTELLIGENCE (AI)-DRIVEN SPECTRUM OPTIMIZATION FRAMEWORK TO ENHANCE QUALITY OF SERVICE IN NIGERIAN MOBILE NETWORKS

The telecommunications industry in Nigeria has been characterised by a boom in mobile subscriptions and internet penetration within the country in the last ten years, but Quality of Service (QoS) issues such as network jamming, high latency, and call drops have been persistently experienced. Much of these predicaments can be blamed on the poor and inflexible allocation policy with regard to spectrum. This paper introduces an Artificial Intelligence (AI)-based spectrum optimization model that can be utilised to improve the quality of service of mobile networks in Nigeria using Deep Reinforcement Learning (DRL), that is, a Deep Q-Network (DQN) model. The representative of the network performance data was the Nigerian Communications Commission (NCC), and the data provided by it covered such parameters as spectrum utilisation, latency, throughput, and rates of call drops. The model was trained and tested in simulated network environments of different traffic conditions. Results show that the proposed DQN-based system outperformed conventional spectrum allocation methods such as Fixed Allocation (FA) and Round Robin (RR) achieving a 35% improvement in throughput, 28% reduction in latency, 23% increase in spectrum utilization, and 40% decrease in call drop rate. These results prove that the model can optimally dynamically use the spectrum and improve network performance. This paper concludes that the incorporation of AI into spectrum management would go a long way in enhancing QoS, efficient 5G deployment, and the digital transformation agenda in Nigeria.