Compressive Sampling for Efficient Astrophysical Signals Digitizing: From Compressibility Study to Data Recovery

The design of a new digital radio receiver for radio astronomical observations in outer space is challenged with energy and bandwidth constraints. This paper proposes a new solution to reduce the number of samples acquired under the Shannon-Nyquist limit while retaining the relevant information of the signal. For this, it proposes to exploit the sparsity of the signal by using a compressive sampling process (also called Compressed Sensing (CS)) at the Analog-to-Digital Converter (ADC) to reduce the amount of data acquired and the energy consumption. As an example of an astrophysical signal, we have analyzed a real Jovian signal within a bandwidth of 40MHz. We have demonstrated that its best sparsity is in the frequency domain with a sparsity level of at least 10% and we have chosen, through a literature review, the Non-Uniform Sampler (NUS) as the receiver architecture. A method for evaluating the reconstruction of the Jovian signal is implemented to assess the impact of CS compression on the relevant information and to calibrate the detection threshold. Through extensive numerical simulations, and by using Orthogonal Matching Pursuit (OMP) as the reconstruction algorithm, we have shown that the Jovian signal could be sensed by taking only 20% of samples at random, while still recovering the relevant information.