Performance metrics




Energy detection

- at high SNRs, the performance is good

- at low SNRs, the performance is unreliable

- affected by noise uncertainty

- Low implementation complexity

- convergence is reach by collecting higher number of samples

- orior information of PU signal isn’t required

- inappropriate for spread spectrum signals

- cannot differentiate a PU signal from other signal sources

Feature detection

- the performance is good at all SNRs

- medium complexity

- convergence requires small number of samples

- require partial knowledge of PU signal

- robust against noise uncertainty and interference

- differentiate PU signal among different types of signal source

Matched filter and coherent detection

- best performance at all SNRs

- high complexity

- convergence requires fewest number of samples

- require precise prior information PU signals

Covariance-based detection

- detection accuracy is high

- low computational complexity

- uncorrelated PU signals degrade detection Performance.

- blind detection