Towards Green Future Cellular Network in Nigeria: Artificial Intelligence Approach

Abstract

For the purpose of lowering energy costs and minimizing the use of fossil fuels, the information sector has traditionally targeted green communications. Without a doubt, the number of linked terminals and the amount of network equipment will continue to grow dramatically in the current 5G and next 6G eras, driving up energy costs. It is becoming more and more crucial to promote the advancement of green communications. But there is no denying that 6G will come with a host of new and more demanding specifications for intelligence, security, flexibility, and Quality of Service (QoS), all of which will make it harder to increase energy efficiency. Additionally, the dynamic energy harvesting process which is expected to be widely used in 6G makes network administration and power control even more difficult. Artificial Intelligence (AI) has been widely regarded as the sole way to handle these issues and minimize the need for human intervention. In order to reduce energy consumption, increase energy efficiency, and control energy harvesting in many communication settings, academia and industry have undertaken a great deal of research. The primary factors for green communications are discussed in this study, along with a review of relevant studies on AI-based green communications. This work focus is on the application of AI approaches to network management and energy efficiency enhancement as we move toward a greener future. The ways in which advanced Deep Learning (DL) and other Machine Learning (ML) approaches can work in tandem with traditional AI techniques and mathematical models to lower algorithm complexity and increase accuracy rates in future communication beyond 5G was examined. Lastly, the current concerns and unresolved research questions related to AI Techniques for future green communication was presented. Relative research towards green cellular network communication (CNC) was presented and it shows that Heuristic algorithms are widely used. Both flexibility and efficiency can be increased, using Heuristic algorithms and machine learning (ML) together. Also, Reinforcement Learning (RL) and Deep reinforcement learning (DRL) approaches helps to achieve the best policy for resource allocation and power control. However, the training process is challenged by the extraordinarily large action space resulted from nature of the metrics taken into consideration.