Cognitive space-time processing with known channel correlations: A nonlinear programming approach

We consider a spectrum sharing based cognitive radio network where a secondary multiple-input multiple-output (MIMO) communication link coexists with a primary user by opportunistically utilizing the transmit spectrum originally allocated to the primary user (PU). The secondary link employs orthogonal space-time block coding (OSTBC) at the transmitter and maximum likelihood (ML) decoding at the receiver. For this system, we focus on designing linear precoders assuming the knowledge of the transmit antenna fading correlations at the secondary user transmitter (ST) for both the links between PU receiver (PR) and ST (ST-PR), and SU receiver (SR) and ST (ST-SR). Given a limited power budget of ST, the precoder is designed to minimize the minimum-distance average pair-wise error probability (PEP) at SR subject to the average interference-power constraint at PR. The original optimization problem is non-convex and we propose to solve this problem using a nonlinear programming method called the augmented Lagrangian algorithm. Simulation results show that this scheme performs better than previously designed suboptimal schemes at all signal to noise ratio (SNR) levels.