Survey on limitations, applications and challenges for machine learning aided hybrid FSO/RF systems under fog and smog influence

Abstract

This paper provides a concise overview of hybrid Free Space Optical (FSO)/Radio Frequency (RF) systems operating under fog and smog channels. With the rapid growth of wireless communication and the emergence of the Internet of Things (IoT) have brought about unprecedented connectivity and transformed various aspects of our daily lives. However, certain environmental conditions pose significant challenges to wireless communication systems like FSO and RF technologies. Fog and smog with its dense moisture-laden atmosphere and presence of particulate matter cause absorption and scattering of light, leading to reduced link quality and limited range for FSO communication system. Whereas, RF systems are although less affected by fog/smog as compared to rain but still have inherent limitations in terms of bandwidth and capacity. To overcome these challenges and ensure reliable communication, researchers have turned their attention to hybrid FSO/RF systems which can combine the strengths of FSO and RF technologies and mitigate the limitations of each technology. This work summarizes and explores the hybrid system architecture and machine learning techniques used to enhance the performance and adaptability of wireless communication systems. Additionally, we present the existing experimental research on hybrid FSO/RF systems, performance evaluation metrics, and future research directions to pave the way for further advancement. Abbreviation FSO: Free Space Optical; RF: Radio Frequency; IoT: Internet-of-Things; OS: Operating Systems; OWC: Optical Wireless Communication; LOS: Line-of-Sight; LPF: Low Pass Filter; AM: Amplitude Modulation; FM: Frequency Modulation; PSK:Phase Shift key; QAM: Quadrature Amplitude Modulation; OFDM: OrthogonalFrequency Division Multiplexing; LF: Low frequency; M-L: Medium frequency; UHF:Ultra High Frequency; SHF: Super High Frequency; EHF: Extremely High frequency; PM: Phase Modulation; QPSK: Quadrature Phase Shift Key; DSSS: Direct SequenceSpread Spectrum; FHSS: Frequency Hopping Spread Spectrum; LAN: Local AreaAccess; D/A: Digital to Analog; LO: Local Oscillator; PA: Power amplifier; A/D:Analog to Digital; LNA: Low Noise Amplifier; FEC: Forward error correction; APC: Automatic Power Control; RSSI: Received Signal Strength Indicator; ML: Machine Learning; RL: Reinforcement Learning; DL: Deep Learning; LR:Logistic Regression; DT: Decision Tree; SVM: Support Vector Machine; KNN:K-Means Neighbour; PCA: Principal Component Analysis; QL: Q-Learning; DQN:Deep Q-NEtwork; R-F: Random Forest; GB: Gradient Boosting; NA: NaiveBayes; ANN: Artificial Neural Network; DTA: Decision Tree with AdaBoost; MIMO: Multiple-Input Multiple-Output; GEO: Geostationary Orbit; NIR: NearInfra-Red; VLC: Visible Light Communication; OP: Optical Communication; BER: Bit Error Rate; ECC: Ergodic Channel Capacity; SNR: Signal-to-NoiseRatio; AO: Adaptive Optics; ATP: Acquisition- Tracking and Pointing; LEO:Low-Altitude Earth Orbit; UAVs: Unmanned Aerial Vehicles; NTN: Non-TerrestrialNetwork; RSSI: Received Signal Strength Indicator; FSPL: Free Space Path Loss; ITU: International Telecommunication Union; GANs: Generative AdversarialNetworks.