One of the most difficult hurdles in the implementation of cognitive radios 
for spectrum sharing applications is that of performing adequate spectrum 
sensing.  For these applications, harmful interference may result if 
sensing is not reliable, efficient, and timely.  In this talk, we introduce 
and analyze algorithms for modulation classification and signal-to-noise 
ratio (SNR) estimation that further the development of this critical 
sensing stage.  The first of these algorithms is an asynchronous and 
non-coherent modulation classifier for PSK/QAM modulated signals in 
flat-fading channels.  This classifier uses low-complexity estimators for 
the unknown time delay, channel gain, phase, and noise power which are 
blind to the modulation format of the received signal.  Using these 
low-complexity estimators, we then present an asynchronous SNR estimator 
for PSK/QAM modulated signals in flat-fading channels which requires no 
a priori knowledge of the modulation format and channel parameters.