To reduce the atmospheric turbulence-induced power loss, an Alex Net-based convolutional neural network(CNN) for wave-front aberration compensation is experimentally investigated for free-space optical(FSO) communicat...To reduce the atmospheric turbulence-induced power loss, an Alex Net-based convolutional neural network(CNN) for wave-front aberration compensation is experimentally investigated for free-space optical(FSO) communication systems with standard single mode fiber-pigtailed photodiodes. The wavefront aberration is statistically constructed to mimic the received light beams with the Zernike mode-based theory for the Kolmogorov turbulence. By analyzing impacts of CNN structures, quantization resolution/noise, and mode count on the power penalty, the Alex Net-based CNN with 8 bit resolution is identified for experimental study. Experimental results indicate that the average power penalty decreases to 1.8 d B from 12.4 d B in the strong turbulence.展开更多
基金This work was supported by the National Natural Science Foundation of China(Nos.61971394 and 61631018)the Key Research Program of Frontier Sciences of CAS(No.QYZDYSSW-JSC003)the Fundamental Research Funds for the Central Universities(No.WK3500000006).
文摘To reduce the atmospheric turbulence-induced power loss, an Alex Net-based convolutional neural network(CNN) for wave-front aberration compensation is experimentally investigated for free-space optical(FSO) communication systems with standard single mode fiber-pigtailed photodiodes. The wavefront aberration is statistically constructed to mimic the received light beams with the Zernike mode-based theory for the Kolmogorov turbulence. By analyzing impacts of CNN structures, quantization resolution/noise, and mode count on the power penalty, the Alex Net-based CNN with 8 bit resolution is identified for experimental study. Experimental results indicate that the average power penalty decreases to 1.8 d B from 12.4 d B in the strong turbulence.
