Skip to Main Content
The problem of robust joint nonlinear transceiver design in a multiuser MIMO interfering Cognitive Radio Network (CR-Net) is studied in this paper. Because of the nonlinear nature of the precoding and equalizing schemes, the transmit or receive parties need to be fed back with some information from the other peers, which gives rise to the concept of the cooperation between the transmission peers. This network is a general multi-party network having all the links interfering with each others. It is also assumed that the Channel State Information (CSI) in this network for all the relevant channels is imperfectly known. The CSI is subject to a Stochastic Error (SE) model- based uncertainty. The design procedure is aiming to satisfy the average performance measures. The chosen performance measure is to minimize the sum Mean Square Error (MSE) of the symbol detection for all the cognitive links of the system while satisfying both Secondary Users' (SU) transmit power and Primary Users' (PU) interfering power constraints. This problem is not jointly convex in the design variables and also has infinitely many constraints, so to overcome these, a suboptimal iterative method is proposed. It is shown that for this SE model, the aforementioned problem is a Second Order Cone Program (SOCP). Finally numerical simulations are provided to show the performance of the proposed methods.
Date of Conference: 6-9 Sept. 2010