Block QR decomposition and near-optimal ordering in intercell cooperative multiple-input multiple-output-orthogonal frequency division multiplexing

Here, the authors investigate dirty paper coding (DPC) in intercell cooperative multiple-input multiple-output-orthogonal frequency division multiplexing (MIMO-OFDM). Based on the multidimensional structure of intercell cooperative MIMO-OFDM, the DPC model has been modified by introducing a block QR decomposition (BQRD) algorithm which transforms the multiuser channel to a block lower triangular structure. The overall error performance of the proposed BQRD-based DPC is dominated by the error performance of the last precoded user. To improve the last user´s performance, it was, first, shown that, in a system with U active users, the channel gain matrix remains unchanged for any user acting as the last user among all (U-1)! possible permutations of the other users. Then, a greedy algorithm that directly computes U channel gain matrices and obtains a near-optimal precoding order for the last user is proposed. The proposed greedy ordering scheme reduces the computational complexity by a factor of U!/2 compared with the brute force search. Our simulation results confirm that the performance of the greedy ordering scheme approaches that of the brute force search. Furthermore, the authors show that the proposed BQRD-based DPC expands the multiuser rate region and increases the sum rate over block diagonalisation based zero forcing method.