<![CDATA[In recent years, dynamic and complex development in wireless communication in network models or environments led to more tedious and complicated resource management issues (i.e., power allocation and base station switching (BSS)). Conventional solutions often suffer from delays and degraded network service quality. Due to the ability of machine learning in analyzing huge volumes of data and automatically adapt to environmental changes, it emerges as a highly sought-after technique. In this work, we propose a machine learning approach based on feed-forward neural network (FFNN) to predict the active BS sets and estimate the power allocation to each user equipment (UE) within the active BSs for energy-efficiency (EE) maximization of a coordinated multi-point (CoMP-enabled) cellular system with hybrid-powered transmitting nodes in a HetNet-based architecture. By training the neural network model efficiently using a regression-based supervised learning technique that employs various backpropagation algorithms, almost similar EE performance (less than 5% difference) can be achieved with significantly reduced computational complexity and delay compared to the traditional methods, such as the well-known dual decomposition and brute force techniques. The effects of various hyper parameters and back-propagation algorithms are also investigated. Our results demonstrate that the proposed framework is a promising solution for establishing a fully green and intelligent network.]]>
Copyrights © 2024
