Dual Branch Compressive Sensing for analog signal

The framework of Analog-to-Information Converter (AIC) enables the feasibility of applying Compressive Sensing (CS) theory into practice. Based on the AIC structure, sparse signals are compressed at sub-Nyquist sampling rate, which dramatically reduces the hardware pressure of A/D chip. However, not fully exploiting the inner correlation of complex frequency, AIC structure does not work well in non-ideal environment or some strict conditions. In this paper, we propose a Dual Branch Compressive Sensing (DBCS) framework in light of the symmetry of Discrete Fourier Transform (DFT). By recovering the real and imaginary parts of complex-valued frequency respectively, we find a way to remove redundancy in reconstructing a complex frequency. Numerical experiments have shown the DBCS structure improves the performance of anti-noise and compressibility remarkably.