Low-latency implementation of CORDIC-based sorted QR decomposition for high-speed MIMO-OFDM system

In this paper, we propose an low-latency coordinate rotation digital computer(CORDIC)-based sorted QR decomposition (SQRD) for high-speed Multiple-Input and Multiple-Output (MIMO) systems. Sorted QRD has been widely adopted to MIMO signal processing to improves the system performance. However, its iterative sorting process after every column operation hinders parallelism and leads to high latency. The proposed CORDIC-based sorted QRD modifies real-value decomposition (RVD) matrix to reduce latency and hardware complexity. It jointly performs sorting processes and CORDIC Givens rotations with adjacent symmetric columns and reduces the numbers of required sorting process and CORDIC stage. The numerical results for a 4 × 4 MIMO show the proposed algorithm reduces the number of CORDIC rotation by 20% in comparison with the conventional sorted QRD. Also, the proposed algorithm requires only 57% of the number of processing stage. The proposed algorithm has been implemented with standard 65 nm cell libraries and achieves reduced latency and complexity compared to the existing results.