Adaptive Autoregressive Prediction Method for Deep-Space Channel Using Kalman Filter

Aiming at improving the signal strength fluctuation phenomena in the deep-space channel affected by solar scintillation during superior conjunction period, a Kalman based adaptive Autoregressive (AR) prediction method for deep-space channel is proposed to predict the channel fading envelope. First, a Rician fading deep-space channel model is established, then the computational methods to channel parameters are dramatically simplified, which are solar scintillation index and Doppler bandwidth respectively, and finally Kalman recursion equations and linear interpolation are applied to modify the prediction results from AR algorithm, realizing the long-term adaptive prediction of the deep-space channel envelope at Ka-band and X-band. Simulation results show that: Compared to the traditional AR algorithm, the proposed method possesses better adaptive characteristics and stronger resistance to multipath fading capability, in the case of weak solar scintillation, the proposed method requires lower order, shorter training data window length, in the case of strong solar scintillation, the mean square error (MSE) of the prediction results can also be reduced by approximately one order.