Efficient detection of FSK-signals based on cyclic statistics

Cognitive radios represent a powerful technology for improving the utilization of radio spectrum resources. Detection of low signal-to-noise ratio (SNR) signals with relaxed a priori information on the signal parameters is of high importance for cognitive radios. This paper proposes a detection algorithm based on the first-order cyclostationarity (or cyclic statistics) for frequency shift keying (FSK) signals that only requires partial knowledge on the signal bandwidth, carrier frequency and symbol rate. The results of both theoretical asymptotic analysis and computer simulations are presented. It is shown that the proposed algorithm performs well at low SNRs.