Deep reinforcement learning based load balancing scheme in dense cellular network using RoF technology

In a dense cellular network, the small cell size and limited frequency make it hard to control the traffic, and hence, there is a necessity for the transmission points to know how much traffic they can handle. To fix this problem in the network, this study suggests a Load Balancing (LB) scheme based on Reinforcement Learning (RL) named DRL-LB adopting a Deep Deterministic Policy Gradient (DDPG) RL approach for a dense cellular network utilizing the RoF technologies. The DRL-LB technique is based on self-exploration in the continuous action space to speed up the execution process. The SNR of the dense network has been taken into account to increase the network spectral efficiency concerning the number of users. The number of users per base station satisfying the minimum SNR value acts as the LB constraints in the scheme. The result analysis shows that it can achieve the required 10 dB of SNR value with 1.6 bits/s/Hz spectral efficiency. It attains a higher spectral efficiency and rewards around 78% compared to the non-LB approach in the scheme. Furthermore, the simulation process also depicts that DRL-LB is 73% more efficient in running time.

Text 125589.pdf - Published Version Available under License Creative Commons Attribution Non-commercial. Download (1MB) Official URL or Download Paper: https://jcoms.fesb.unist.hr/10.24138/jcomss-2025-0...

Actions (login required)

View Item