Diversified Compressed Spectrum Sensing for Recovery Noise Reduction

Abstract

We propose a method to reduce the spectrum noise when compressed sensing (CS) is applied to spectrum sensing. Since CS is susceptible to noise, the quality of the recovered spectrum using CS can be significantly degraded if the measurements are contaminated with noise. This will be particularly problematic in case of detecting weak signals. In this paper, a method which exploits the diversity of CS measurements is introduced to reduce the noise of CS recovered spectrum. Diversity gain is extracted from measurements using only a single physical sensor, but with virtual parallel branches. The results show that the noise is reduced sufficiently to detect weak signals using our method.