A Robust Spectrum Sensing Method Based on Maximum Cyclic Autocorrelation Selection for Dynamic Spectrum Access

Kazushi MURAOKA  Masayuki ARIYOSHI  Takeo FUJII  

IEICE TRANSACTIONS on Communications   Vol.E92-B   No.12   pp.3635-3643
Publication Date: 2009/12/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.E92.B.3635
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Dynamic Spectrum Access)
Category: Spectrum Sensing
cognitive radio,  dynamic spectrum access (DSA),  spectrum sensing,  cyclostationary feature detection (CFD),  maximum cyclic autocorrelation selection (MCAS),  

Full Text: PDF(796.2KB)>>
Buy this Article

Spectrum sensing is an important function for dynamic spectrum access (DSA) type cognitive radio systems to detect opportunities for sharing the spectrum with a primary system. The key requirements for spectrum sensing are stability in controlling the probability of false alarm as well as detection performance of the primary signals. However, false alarms can be triggered by noise uncertainty at the secondary devices or unknown interference signals from other secondary systems in realistic radio environments. This paper proposes a robust spectrum sensing method against such uncertainties; it is a kind of cyclostationary feature detection (CFD) approaches. Our proposed method, referred to as maximum cyclic autocorrelation selection (MCAS), compares the peak and non-peak values of the cyclic autocorrelation function (CAF) to detect primary signals, where the non-peak value is the CAF value calculated at cyclic frequencies between the peaks. In MCAS, the desired probability of false alarm can be obtained by setting the number of the non-peak values. In addition, the multiple peak values are combined in MCAS to obtain noise reduction effect and coherent combining gain. Through computer simulations, we show that MCAS can control the probability of false alarm under the condition of noise uncertainty and interference. Furthermore, our method achieves better performance with much less computational complexity in comparison to conventional CFD methods.