Joint Rapid Spectrum Scanning and Signal Feature Recognition Scheme Using Compressed Sensing and Cyclostationary Technologies

Abstract

This paper proposes a joint rapid spectrum scanning and signal feature recognition scheme by using both compressed sensing (CS) and cyclostationary technologies to solve the paradox of rapid spectrum scanning and accurate signal feature recognition. First, a new system architecture for signal feature recognition is designed based on rapid spectrum scanning results to decrease the times of proposed signal feature recognition scheme. Moreover, an improved CS technology is brought forward to accelerate the sensing speed without reconstructing received signals for a rapid spectrum scanning. And a tunable compression gain is proposed to reduce both computation complexity and sampling rate based on the difference of modulation mode and symbol rate for various signals. To further reduce the effect of noise on the modulation classification performance, a novel noise reduction scheme is proposed using the cyclostationary technology. Results prove that proposed scheme can achieve both rapid spectrum scanning and accurate signal feature recognition simultaneously. Furthermore, it can reduce the sampling rate for CS over 30% and achieves a signal detection gain of 2–3 dB with signal to noise ratio constraints.