Parameter Optimization for Conventional Soft Frequency Reuse in Multi Cell Networks using Extreme Learning Machine and Genetic Algorithm

Abstract

Soft Frequency reuse (SFR) has been used as a method to reduce the inter-cell interference between cell edge user and cell center users. With a proper adjustment of the system, it can gives us the best throughput of the SFR. The objective of this study is to optimize the three parameters of the SFR obtained from Taguchi model. These three parameters consist of power ratios of the base station (β), concentration of cell-edge user and user distribution of sub-subcarriers. Firstly, the SFR data is modeled using Extreme Learning Machine (ELM). The model is then validated using the unseen samples based on the mean square error (MSE) criterion. Secondly, the best model with the lowest MSE is then used for the parameters optimization by using genetic algorithm (GA) in order to obtain the highest throughput. In the series of experiments, the obtained results offer the highest throughput with reliable and consistent selected parameters. In this paper, the combination ELM-GA technique has been proven to increase the throughput of the system.