Cognitive beamforming made practical: Effective interference channel and learning-throughput tradeoff

This paper studies the transmission strategy for a cognitive radio (CR) that operates under spectrum sharing with a primary radio (PR). It is assumed that the CR transmitter is equipped with multi-antennas and thereby transmit beamforming and power control are jointly deployed to balance between the interference avoidance at the PR terminals and the throughput maximization of the CR link. We name such an operation as cognitive beamforming (CB). Unlike the prior study on CB that assumes perfect knowledge on the interference channels over which the CR transmitter interferes with the PR terminals, in this paper we remove this assumption and propose a practical CB scheme by utilizing a new idea of effective interference channel, which can be efficiently learned/estimated at the CR transmitter from the received PR signals. Interestingly, it is shown that the proposed CB scheme based on the effective interference channel achieves a capacity gain for the CR over the conventional scheme based on the exact channel knowledge, when the PR terminals are also equipped with multi-antennas but do not operate in a full spatialmultiplexing mode. Furthermore, due to the finite learning time for the effective interference channel, we show that there exists a general learning-throughput tradeoff associated with the practical CB, for which we determine the optimal learning time to maximize the CR link throughput.