Machine Learning in Energy Efficiency Optimization

This chapter discusses machine learning (ML) as a means to improve energy efficiency (EE) of wireless networks. In this sense, it reviews the most common ML approaches with focus on the maximization of EE. The chapter first gives a brief definition of self-organizing networks (SONs) and some related solutions involving ML algorithms in the context of EE. SONs can be divided into three main branches: self-configuration, self-optimization, and self-healing, together denoted as self-x functions. The chapter illustrates some applications of SONs in cellular networks, highlighting the three major self-x branches and the common associated use cases. In this context, ML techniques can certainly improve a SON, allowing the network to adapt by observing its current status, and use such experience to adjust parameters in future actions. The chapter also presents an overview of ML techniques applied to more specific topics such as resource allocation, traffic prediction, and cognitive radio networks.