Bayesian Inference for Multiple Antenna Cognitive Receivers

In this paper, we provide a Bayesian learning process for cognitive devices. In particular we focus on the case of signal detection as an explanatory example to the learning framework. Under any prior state of knowledge on the communication channel, an information theoretic criterion is presented to decide if informative data is present in a noisy wireless MIMO communication. We detail the particular cases of knowledge, or absence of knowledge at the receiver, of (i) the number of transmit antennas and (ii) the effective noise power. The provided method is instrumental to embed intelligence into the wireless device and gives birth to a novel Bayesian signal detector which is compared to the classical power detector. Simulations corroborate the theoretical results and quantify the gain achieved by the proposed Bayesian framework.