Separating the Wheat from the Chaff: Sensing Wireless Microphones in TVWS
This paper summarizes our attempts to establish a systematic approach that overcomes a key difficulty in sensing wireless microphone signals, namely, the inability for most existing detection methods to effectively distinguish between a wireless microphone signal and a sinusoidal continuous wave (CW). Such an inability has led to an excessively high false alarm rate and thus severely limited the utility of sensing-based cognitive transmission in the TV white space (TVWS) spectrum. Having recognized the root of the difficulty, we propose two potential solutions. The first solution focuses on the periodogram as an estimate of the power spectral density (PSD), utilizing the property that a CW has a line spectral component while a wireless microphone signal has a slightly dispersed PSD. In that approach, we formulate the resulting decision model as an one-sided test for Gaussian vectors, based on Kullback-Leibler distance type of decision statistics. The second solution goes beyond the PSD and looks into the spectral correlation function (SCF), proposing an augmented SCF that is capable of revealing more features in the cycle frequency domain compared with the conventional SCF. Thus the augmented SCF exhibits the key difference between CW and wireless microphone signals. Both simulation results and experimental validation results indicate that the two proposed solutions are promising for sensing wireless microphones in TVWS.