Spectrum sensors for cognitive radio are expected
to deploy multiple antennas in order to overcome the noise
uncertainty problem and minimize the effects of small-scale
fading. Despite the requirement that these sensors must detect
wireless microphone (WM) signals, works in the literature have
focused either on general purpose multiantenna detectors, or
single antenna WM detectors. We exploit the spatial structure
and particular properties of WM waveforms to derive four
multiantenna detectors for WM signals with different performance/
complexity tradeoffs. These detectors are based on the
generalized likelihood ratio test, which is derived under several
signal models exploiting either the fact that the bandwidth of
a WM signal never exceeds 200 kHz, the property that these
signals have a constant magnitude, or both.

The proposed detectors do not require synchronization with
the WM signal and are robust to the noise uncertainty problem as
well as to small-scale fading. Using the simulation guidelines from
the IEEE 802.22 standard, the novel multiantenna WM detectors
are shown to outperform previous schemes, thus demonstrating
the advantages of exploiting spatial correlation along with WM
signal structure.