KTH - Royal Institute of Technology
A multi-antenna cognitive radio network, with a single pair of primary users and a secondary broadcast channel, is considered. Under perfect Channel State Information (CSI), the rate-optimal strategy for the primary link is waterfilling, resulting in possibly unused dimensions. The secondary base station, supplied with perfect and global CSI, employs block diagonalization for interference-free message precoding and opportunistic interference alignment to avoid disturbing primary communication. In the absence of sufficient resources to serve all secondary users, a low complexity adaptive antenna selection strategy is proposed. This scheme constructs, in each timeslot, the set of active antennas by opportunistically reusing information and resources from the preceding selection round.