Nonlinear Multiplier and Deep Learning Based Fixed Threshold On-Off Keying Detection for Free-Space Optical Communications

On-off keying (OOK) signal is affected by the scintillation effect caused by atmospheric turbulence when it is transmitted through atmospheric channel, which results in intensity fluctuation of received signal. This paper proposes nonlinear multiplier (NM) and deep learning (DL) based fixed threshold OOK detection for free-space optical (FSO) communications to compensate the scintillation effect. The strength of the received signal is improved due to the multiplier characteristic of providing different amplification gains according to different signal strengths. Three kinds of nonlinear multipliers are studied in this paper. However, the NM based fixed threshold OOK detection technique is less effective at higher atmospheric turbulence intensities due to the distortion of the OOK signal extinction ratio (ER) caused by the use of highly non-linear compensation. Therefore, an improved Gated Recurrent Unit (GRU) model is used to assist NM. Simulation experiments were conducted at different turbulence intensities. The results showed that the proposed method outperforms traditional fixed threshold decision (FTD), NM based fixed threshold OOK detection, improved GRU based fixed threshold OOK detection, and adaptive threshold decision (ATD). The atmospheric turbulence scintillation effect is effectively compensated.