Performance of adaptive M-PAM modulation for FSO systems based on end-to-end learning

This paper examines the performance of free space optics (FSO) utilizing M-ary pulse amplitude modulation (M-PAM). We utilize end-to-end learning to optimize the FSO transmitter and maximize data throughput. The main goal is to improve system performance using artificial neural networks for automatically changing modulation schemes to maximize communication speed subject to the possible transmitted power instead of channel state information (CSI). Nevertheless, inadequate or outdated CSI may pose challenges in extracting the anticipated theoretical gain from adaptive systems. The proposed system is designed to analyze the data obtained from the receiver, such as bit-error-rate (BER), noise, and atmospheric conditions, to select the highest modulation order. End-to-end machine learning allows the transmitter to tune modulation order for M-PAM depending on the required transmitted power, the desired BER, and the FSO channel status. Simultansiually, the receiver uses a channel estimator to estimate channel frequency response in the presence of atmospheric turbulence, and the equalizer can compensate for the signal distortions. Simulation results confirm the validity of the proposed system by comparing it with traditional adaptive M-PAM.