A low complexity algorithm and architecture for MIMO detection without QR decomposition

This paper presents a novel low-complexity, hard output multiple-input, multiple-output (MIMO) detection algorithm and architecture for LTE and WiFi applications. Our method provides a well-defined tradeoff between computational complexity and performance. The proposed algorithm does not require QR decomposition as a preprocessing step, as is commonly needed in sphere detection (SD). Simulation results for 3×3 and 4×4 MIMO configurations with both 16 and 64 QAM constellations show that the proposed algorithm uses a much smaller number of Euclidean distance (ED) calculations than other methods while having only a 0.5 dB loss compared to maximum likelihood detection. A 3×3 MIMO system with a 16 QAM detector architecture is also described, and the associated latency and hardware costs are given.