As the critical component of the deep space observation network, the receiver system observes the radio signals from satellite or spacecrafts for navigation, ranging, planetary atmospheric scientific research and etc. The current receiver systems have the problems of weak universality and low efficiency on signal processing speed. Therefore, the development of receiver system with high accuracy, strong robustness and generality is meaningful and necessary.
In the paper, for deep space exploration applications, the signal processing algorithms with high accuracy, strong robustness is studied and the receiver system with generality is proposed.
The main contribution of the paper is shown as follows. Firstly, on the research and implementation on the narrowband signal parameter estimation algorithm, several basic parameter estimation algorithms are improved. Then, based on the comparison results, the adaptive multi-stage parameter estimation algorithm is proposed. During the two-stages of parameter estimations, the proposed algorithm adjusts the signal detection window according to the current Doppler information. In this way, the interference of random noise is suppressed and the system robustness is enhanced. The experiment based on the real satellite data observations demonstrate that the system keeps a continuous observation results without any breakpoint during the long time of observations. Meanwhile, the high accuracy of the observation data is achieved but has less requirement of the signal SNR. The RMSE for narrowband signal frequency estimation, Doppler rate and phase estimation is about 0.01Hz, 0.02Hz/s and 0.06rad, respectively. Secondly, on the research and implementation of the wideband signal time delay estimations, to solve the wideband signal time delay estimations based on the inaccurate forecast data, a novel algorithm is proposed based on the multiple correlation and 2-D correlation chart analysis. Multiple correlation method is employed to enhance the SNR of the correlation function in the first place. Then, the results are represented on the 2-D contour map to acquire the correlation chart. After the rough delay results got from the chart, the Doppler compensation is applied to calculate the fine estimation results. The real data experiment demonstrates the effectiveness of the proposed method. The accurate time delay estimation results can still be acquired when the forecast data is inaccurate in the wideband signal time delay estimation. The fitting residue of the algorithm is about 0.1rad, which demonstrates the similar accuracy acquired by the estimation with accurate forecast data. On the research and implementation of the points changeable FFT processor, based on the 2-D FFT decomposition algorithm, the proposed FFT processor design decomposes the processor into 1k points parallel FFT sub-module and 32 points changeable sub-nodule. The processor design improves the FFT computation efficiency. The 32K FFT verification experiment demonstrate the maximum processing error for designed processor is less than 1%, the processing speed is nearly 20 times for FFT processor compared to the software processing. Finally, on the research and implementation of intelligent signal processing and general purpose ANFIS processor, a general purpose ANFIS processor for system intelligent signal processing algorithms is presented to improve the system computation efficiency. The universality of the processor is achieved through the design of parameter run-time tunable non-linear membership function generator and reconfigurable arithmetic computation network. The Mackey-Glass chaotic series data experiment demonstrates the learning RMSE for proposed ANFIS processor is 0.03 and the processing time is nearly 20% compared to the software processing implemented on MATLAB. Meanwhile, in the experiment simulates the reference data is off-line for 60 seconds, the proposed hybrid method with designed ANFIS processor suppress most of the accumulation error for open-loop integration phase measurement, which demonstrate the design of the hybird algorithm and ANFIS processor are effective.