A 2×2–8×8 sorted QR decomposition processor for MIMO detection

Due to the growing demand of transmission capacity of the wireless communication system, multiple-input multiple-output orthogonal-frequency-division-multiplexing (MIMO-OFDM) communication system requires more and more MIMO antennas and a large number of OFDM subcarriers. Thus, the QR decomposition becomes one of the computational bottlenecks in the QR-based MIMO detection. The proposed Givens-Rotation-based QR decomposition algorithm features efficient parallel processing with sorting function that resolves the trade-off between the detection performance and hardware utiliztion efficiency. According to the algorithm, we designed and implemented a 2×2∼8×8 QR decomposition processor using TSMC 0.18/μm 1P6M CMOS technology. The throughput of the QR decomposition processor achieves 6.8×10
⁴
SQRD per second, which outperforms other works in the literature after being normalized by the MIMO dimension.