Echo State Network for Turbulence-Induced Fading Channel Prediction in Free-Space Optical Systems
Echo State Network for Turbulence-Induced Fading Channel Prediction in Free-Space Optical Systems
复制标题
DOI：
10.1109/wsce56210.2022.9916027
复制
发表时间：
2022-09
期刊：
影响因子：
--
通讯作者：
T. Nguyen;Hoang D. Le;A. Pham
中科院分区：
文献类型：
--
作者：
T. Nguyen;Hoang D. Le;A. Pham
自由空间光 (FSO) 通信因提供超高数据速率服务而享有盛誉。在时变湍流信道上，由于反馈延迟，信道状态信息（CSI）往往会过时。这可能会导致 FSO 系统的性能显着下降。本文研究了一种在 FSO 湍流引起的衰落信道上准确更新 CSI 的预测方案。我们采用回波状态网络 (ESN) 模型，这是一种计算复杂度低但效率高的循环神经网络 (RNN) 形式，用于 FSO 系统中的湍流通道预测。我们使用测量的 FSO 通道数据验证预测模型。仿真结果表明ESN模型在FSO系统中具有良好的预测性能。此外，我们还通过将 ESN 模型与传统的 CSI 预测模型在复杂性和准确性性能方面进行比较，强调了 ESN 模型的有效性。
Free-space optical (FSO) communication has established its reputation for delivering ultra-high data-rate services. Over the time-varying turbulence channels, the channel state information (CSI) tends to be outdated due to the feedback delay. This may lead to considerable performance deterioration in FSO systems. This paper studies a prediction scheme for accurately updating CSI over FSO turbulence-induced fading channels. We employ the echo state network (ESN) model, a form of recurrent neural network (RNN) with low computational complexity yet high efficiency, for turbulence channel prediction in FSO systems. We verify the prediction model using measured FSO channel data. Simulation results indicate that the ESN model has excellent prediction performance in FSO systems. Furthermore, we also highlight the effectiveness of the ESN model by comparing it with conventional CSI prediction models in terms of complexity and accuracy performance.