斯太尔率
自适应光学
波前
梯度下降
计算机科学
自由空间光通信
光学
变形镜
随机梯度下降算法
光通信
物理
人工智能
人工神经网络
作者
Zhaokun Li,Tao Shang,Xiongchao Liu,Peiheng Qian,Yinling Zhang
标识
DOI:10.1016/j.optcom.2023.129268
摘要
The stochastic parallel gradient descent (SPGD) method has been deeply explored to correct atmospheric turbulence in free-space optical communication. However, SPGD is usually trapped in the problems of sub-optimization and long-time iteration because the controlling signals can be updated merely by the local gradient derived from the Strehl ratio (SR). This research proposes an advanced multi-feedback SPGD (AMF-SPGD) method to strengthen the wavefront aberration correction performance. With the aid of multiplexed computer-generated hologram (CGH), a spot array will be captured thus a corresponding SR array is subsequently obtained, which provides the multi-feedback suggesting the more abundant optimized information. By adequately integrating the global and local gradient, AMF-SPGD could timely adjust the search gradients and achieve the optimum with a greater probability. The system principle and pipeline of the AMF-SPGD will be illustrated in detail. Subsequently, extensive simulations will be conducted to verify the significant upgrade in performance, including the search speed, SR and bit error rate (BER).
科研通智能强力驱动
Strongly Powered by AbleSci AI