HYBRID FIREFLY AND PARTICLE SWARM OPTIMIZATION ALGORITHM FOR DESIGN OF NON-LINEAR CHANNEL EQUALIZER
Artificial Neural Network (ANN) equalizers have indeed proven to be effective tools for mitigating Inter-Symbol Interference (ISI) resulting from distortions introduced by the channel. The purpose of this work is to propose a hybrid Firefly and Particle Swarm Optimization technique (FFPSO) combined...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
University of Kragujevac
2025-03-01
|
| Series: | Proceedings on Engineering Sciences |
| Subjects: | |
| Online Access: | https://pesjournal.net/journal/v7-n1/59.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Artificial Neural Network (ANN) equalizers have indeed proven to be effective tools for mitigating Inter-Symbol Interference (ISI) resulting from distortions introduced by the channel. The purpose of this work is to propose a hybrid Firefly and Particle Swarm Optimization technique (FFPSO) combined using Legendre Neural Networks (LeNN) for the design of channel equalizer. This optimum solution has been modelled using the FFA algorithm, which is driven mainly by light intensity attraction. It is possible that the FFA algorithm will delay achieving a global optimization solution because it relies on arbitrary search directions. This hybrid technique combines the concepts of both the algorithms to create fresh population. In this research, an innovative straegy plan for LeNN equalisers utilising FF-PSO is proposed. Higher exploitation and exploration capabilities, as well as an improved ability to escape from local minima, are features of the suggested training plan. Furthermore, we compare the features of FFPSO with the classical features of FF and PSO. Legendre Neural Network (LeNN) classifiers utilize the FFPSO feature values as inputs. The functioning of the equalisation of the FF-PSO is presented through the simulation of concerned channels, and the outcomes have been compared with those of recently developed and well appreciated methods. As measured by BER and MSE, the simulation results verify that the suggested training scheme outperforms current metaheuristic algorithms by a significant margin. |
|---|---|
| ISSN: | 2620-2832 2683-4111 |