Experimental detection performance of cyclostationarity-based spectrum sensing

Cyclostationarity-based spectrum sensing exploits the spectral correlation property of the primary user (PU) signal; hence, exhibits relative robustness against noise and interference. Existing studies on this method focus on analytical and simulation approaches. This work focuses on the experimental measurement of the performance of cyclostationarity-based spectrum sensing using a magnitude squared coherence (MSC) function. The MSC is the magnitude square of the function that measures the intensity of coherence of the PU signal spectral components. In this work, MSC is estimated using the FFT Accumulation Method (FAM). We implemented a testbed operating on TV white space, and measured the empirical performance of MSC-based detection over the following: (1) the effect of PU duty cycle, and (2) the effect of time-frequency resolution product (TFRP) parameter of FAM. The measured results show that the experimental performance of MSC-based detector has an experimental performance gain of around 6 dB over simple energy detection scheme. The performance deteriorates with decreasing PU duty cycle, which agrees with existing simulation study. Moreover, the empirical performance also degrades with decreasing TFRP, which agrees with the theoretical reliability condition requiring a TFRP of much greater than unity.