Characterization of a Compressive Sensing Preprocessor for Vector Signal Analysis

In spectrum sensing and wireless communications analysis, signals of interest typically occupy only a few among several possible bands and do so for short-time bursts within a given observation interval. The frequency and time location of these signals may be known only approximately a priori (for instance, the nominal frequency of a wireless channel) or, in general, not accurately enough to set up more detailed measurements. In this paper, a compressive sensing (CS) algorithm is employed to provide accurate preliminary information and suitably preprocessed data for a vector signal analysis algorithm. The CS paradigm exploits sparsity, a feature common to several signals of interest, to allow the design of efficient data acquisition schemes. It is shown that the application of the sensing method called modulated wideband converter allows one to successfully extract specific signal bursts from a record of samples covering a longer time interval and a broader bandwidth. The accuracy of the extraction process is analyzed and the results referring to vector analysis are presented. This provides blind spectrum sensing and signal extraction capabilities that can effectively simplify the time-consuming process of setting up a spectrum analyzer for vector signal analysis.