A QR decomposition aided belief propagation (QR-BP) detector for multiple-input multiple-output (MIMO) systems is proposed. Before the iterative message passing process, linear transformation is performed by using the unitary matrix obtained from the QR decomposition of the original MIMO channel matrix. Then, the original MIMO channel matrix is replaced by an upper triangular channel matrix after the linear transformation and the proposed QR-BP detector operates over the new bipartite graph from the upper triangular channel matrix. Therefore, the number of cycles in the new bipartite graph is significantly decreased when compared with the fully connected bipartite graph from the original MIMO channel matrix. In addition, since the unitary matrix is utilised for the linear transformation, there is no information loss on the likelihoods estimated from noisy observations. Simulation results verify that the proposed QR-BP detector outperforms existing BP detectors and achieves near-optimum performance close to the maximum-likelihood detector in terms of bit error ratio.
All Science Journal Classification (ASJC) codes
- Electrical and Electronic Engineering