Résumé
Energy-efficient scalable soft-output signal detectors are of significant interest in emerging Multiple-Input Multiple-Output (MIMO) wireless communication systems. However, traditional high-performance MIMO detectors consume a rather high amount of power, are typically constraint to one modulation scheme and are not scalable with the number of antennas. Hence, they are not well-suited for future energy-efficient Software Defined Radio (SDR) platforms. This paper presents two energy-efficient scalable MIMO detector architectures: one optimized for high throughput, one for low area. Both architectures support 16-QAM as well as 64-QAM while offering soft-output and near-ML performance. The 2 x 2 high-throughput architecture was implemented in CMOS 65nm technology and subsequently scaled to 4 x 4 and 8x8. The 4 x 4 instance provides up to 300Mbps throughput while consuming only 0.3mm(2) area and 28mW power. The 8 x 8 instance offers a throughput 10 x better than the state-of-the-art while consuming 2/3 less power. Thus, the proposed near-ML Selective Spanning with Fast Enumeration (SSFE) based detector architectures are not only multi-standard capable and scalable, they are also highly efficient.