International Conference on Information and Knowledge Technology

Home / شانزدهمین کنفرانس بین المللی فناوری اطلاعات و دانش

Energy-Saving for User-Centric Dynamic 5G HetNets Using DRL Method

Authors :

Erfan Rasti¹ Mohammad Ali Arami² Abbas Mohammadi³

1- دانشگاه صنعتی امیرکبیر (پلی‌تکنیک تهران) 2- دانشگاه صنعتی امیرکبیر (پلی‌تکنیک تهران) 3- دانشگاه صنعتی امیرکبیر (پلی‌تکنیک تهران)

Keywords :

Energy-Saving،Power Consumption،Energy Efficiency،Reinforcement Learning

Abstract :

In modern communication networks, due to the high density of users and network facilities, reducing energy consumption (EC) has become a vital problem to address. In the past few years, there has been some valuable research on energy-saving methods as a solution for this criterion. This paper demonstrates a user-centric method to maximize energy-saving in conjunction with a deep Q-network (DQN) model aligned with the green cellular network. This approach involves a user-centric connection establishment to activate small base stations (SBSs) using a DQN-controlled switching mechanism, which smooths the transitions. Through the request-step hyperparameter, we developed an anti-abrupt mode transition, minimizing network oscillations. Finally, our achievements have been accurately compared with previous works like genetic algorithm (GA) and particle swarm optimization (PSO), convincingly demonstrating that the DQN method presents a significant improvement regarding EC, power consumption (PC), and time complexity.