Beamspace MIMO for high-dimensional multiuser communication at millimeter-wave frequencies

Millimeter-wave (mm-wave) systems operating from 30-300GHz provide a unique opportunity for meeting the exploding capacity demands on wireless networks. In addition to orders-of-magnitude larger bandwidths, the small wavelengths at mm-wave enable high-dimensional multiple-input multiple-output (MIMO) operation. However, fully exploiting the advantages of mm-wave requires prohibitively high transceiver complexity when using conventional MIMO techniques. In this paper we propose and analyze the sum capacity of several linear, reduced-complexity multiuser MIMO (MU-MIMO) precoders that exploit the concept of beamspace MIMO (B-MIMO) communication - multiplexing data onto orthogonal spatial beams that serve as approximate channel eigenfunctions. Due to quasi-optical propagation at mm-wave, MIMO channels are expected to be low-rank and the low channel rank is manifested in the sparsity of the beamspace channel matrix. This enables near-optimal rank and complexity reduction via the concept of multi-beam selection. We present numerical capacity results that demonstrate the reduced-complexity B-MIMO precoders are able to closely approximate the performance of their full-dimensional counterparts with complexity that tracks the number of mobile stations (MSs). In mm-wave systems, where the number of MSs is expected to be much lower than the system dimension, this enables a considerable reduction in the digital signal processing complexity and in systems equipped with analog beamforming front-ends the hardware complexity is also reduced. Thus, the proposed reduced-complexity multiuser precoders provide a near-optimal route for achieving multi-Gigabit/s sum rates in mm-wave MU-MIMO networks with the lowest transceiver complexity.