An averaged-LLR group detection for higher order MIMO system

Designing an efficient MIMO decoder for higher order MIMO system such as 8×8 is a big challenge. The Group Detection (GD) method has been considered as a good choice because it balances well the trade-off between detection accuracy and computational complexity. This method converts the higher order MIMO channel matrix into several lower order matrices and then decodes these lower order matrices separately. In another aspect, a Multi-User MIMO (MU-MIMO) system commonly requires a pre-coding such as Block Diagonalization (BD) to eliminate the inter-user interference (IUI). In this paper, we show that a MU-MIMO system can reuse the BD pre-coding at the transmitter to perform the GD at the receiver for reducing the complexity. To improve the performance, a Group Selection (GS) method should be used before doing GD. While the optimal GS method achieves high performance, its complexity is too high for real applications. For balancing the complexity-performance trade-off, we propose an Averaged Log Likelihood Ratio Group Detection (A-LLR-GD) in this paper. The packet error rate (PER) and Bit Error Rate (BER) simulation results on IEEE 802.11ac simulator show that our proposed one achieves higher performance with low increase in computational complexity as compared to the conventional works.