Spectrum sensing in time-varying channels using multiple antennas

The problem of detecting a Gaussian signal received with multiple antennas through a time-varying channel and corrupted by noise of unknown power is addressed by deriving the Generalized Likelihood Ratio test. The channel is described by means of a basis expansion model with deterministic unknown coefficients. Since the detection rule requires the knowledge of the Maximum Likelihood estimates of the channel and noise power, which lack closed-form expressions, an Expectation-Maximization algorithm is proposed to carry out such computation. Finally, simulations reveal that the proposed detector successfully exploits spatial correlation and time variation of the channel.