ASG: An Adaptive Savitzky–Golay Method for Channel Estimation in Deep-Space Communications

In deep-space communications, the channel is vulnerable to solar scintillation caused by coronal plasma. Furthermore, the significant Doppler frequency shift introduces rapid variations in the channel, making accurate channel estimation especially challenging. To address this issue, we reformulate deep-space channel estimation as a signal denoising problem and introduce an adaptive Savitzky-Golay (ASG) denoising approach. Specifically, we model the time-varying deep-space channel using a Gaussian Doppler spectrum while accounting for the effect of solar scintillation. Then, we use the least squares (LS) method to obtain a rough estimation and apply the SG filter for denoising. Stein's unbiased risk estimation (SURE) serves as the objective function, and a grid search algorithm is utilized to determine the optimal filtering parameters. Simulation results show that the proposed ASG method can reduce mean squared error (MSE) by up to 98% compared to the traditional LS method. More importantly, the MSE performance of the ASG method is close to that of the optimal minimum mean squared error (MMSE) method, without requiring any prior channel information. Furthermore, ASG shows greater robustness than MMSE under different scintillation conditions, demonstrating its reliability in deep-space communication systems.