Sparse sampling for Software Defined Radio receivers

In this paper we address the problem of designing a back-end for Software Defined Radio (SDR) receiver using only one low-rate Analog-to-Digital Converter (ADC). After the Intermediate Frequency (IF) down-conversion stage, the signals are sampled and all the other demodulation tasks are assigned to a Digital Signal Processor (DSP). In this paper we concentrate on only one Quadrature Amplitude Modulation (QAM) signal. After reduction of the communication chain to a linear model, the issue of proper sampling rate choice is addressed. It is shown how improper sampling rate can lead to a poorly conditioned problem that causes symbol cancelations and poor performances. In the analysis the role of the device noise at the input of the ADC is also addressed by computing autocorrelation and Power Spectral Density (PSD) after undersampling. Some simulations confirm the analysis.