IIUM Repository

A Survey on channel impairment resilience with machine learning in free-space optical communication links

Hemed, Salem and Habaebi, Mohamed Hadi and Islam, Md. Rafiqul and Zabidi, Suriza Ahmad (2026) A Survey on channel impairment resilience with machine learning in free-space optical communication links. Computers and Electrical Engineering, 139 Part B (11). pp. 1-28. ISSN 0045-7906 E-ISSN 1879-0755

[img]
Preview
PDF
Download (7MB) | Preview

Abstract

Free-space optical communication has emerged as a critical enabler for next-generation wireless networks, offering license-free spectrum, inherent security, and ultra-high data rates. However, its performance remains highly vulnerable to atmospheric impairments that severely degrade link reliability. Traditional mitigation techniques often rely on linear assumptions and reactive mechanisms, limiting their effectiveness in dynamic environments. Recently, machine learning and deep learning have introduced a paradigm shift, enabling data-driven solutions capable of capturing nonlinear, time-varying channel behaviors for both proactive and reactive resilience. This survey focuses exclusively on deep learning techniques for mitigating free-space optical channel impairments. We propose a unified taxonomy classifying applications into five domains: channel estimation; prediction; modeling; signal detection and recovery; and intelligent switching in hybrid free-space optical/radiofrequency systems. For each category, we review state-of-the-art architectures, synthesizing their performance metrics and inherent limitations. Given the scarcity of real-world free-space optical data, we critically evaluate existing datasets and the resulting simulation-to-reality gap. Furthermore, we identify that practical deployment remains constrained by limited generalization, reliance on simulated datasets, lack of standardized benchmarks, and hardware-model mismatches. The paper highlights these challenges and outlines future research directions focused on dataset standardization, hardware-aware model design, and adaptive learning frameworks to enable reliable real-world deployment of intelligent free-space optical systems.

Item Type: Article (Journal)
Uncontrolled Keywords: Free-space optics Deep learning Channel impairments Atmospheric turbulence Signal detection Channel estimations Dataset scarcity Explainable AI High-altitude platform station
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication. Including telegraphy, radio, radar, television
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button): Kulliyyah of Engineering > Department of Electrical and Computer Engineering
Depositing User: Dr. Mohamed Hadi Habaebi
Date Deposited: 16 Aug 2026 11:07
Last Update: 16 Aug 2026 11:07
Queue Number: 2026-08-Q4749
URI: http://irep.iium.edu.my/id/eprint/130802

Actions (login required)

View Item View Item