Antenna selection is a multiple-input multiple-output (MIMO) technology, which uses radio frequency (RF) switches to select a good subset of antennas. Antenna selection can alleviate the requirement on the number of RF transceivers, thus being attractive for massive MIMO systems. In massive MIMO antenna selection systems, RF switching architectures need to be carefully considered. In this paper, we examine two switching architectures, i.e., full-array and sub-array. By assuming independent and identically distributed Rayleigh flat fading channels, we use asymptotic theory on order statistics to derive the asymptotic upper capacity bounds of massive MIMO channels with antenna selection for the both switching architectures in the large-scale limit. We also use the derived bounds to further derive the upper bounds of the ergodic achievable spectral efficiency considering the channel state information (CSI) acquisition. It is also showed that the ergodic capacity of sub-array antenna selection system scales no faster than double logarithmic rate. In addition, optimal antenna selection algorithms based on branch-and-bound are proposed for both switching architectures. Our results show that the derived asymptotic bounds are effective and also apply to the finite-dimensional MIMO. The CSI acquisition is one of the main limits for the massive MIMO antenna selection systems in the time-variant channels. The proposed optimal antenna selection algorithms are much faster than the exhaustive-search-based antenna selection, e.g., 1000 × speedup observed in the large-scale system. Interestingly, the full-array and sub-array systems have very close performance, which is validated by their exact capacities and their close upper bounds on capacity.
4 October 2019 in IEEE Transactions on Antennas and Propagation
The use of massive Multiple-Input Multiple-Output (MIMO) systems is becoming progressively feasible, as well as progressively important, as the carrier frequency increases. Terahertz (THz) frequencies lead to smaller-sized RF components including antennas, so that even antenna arrays with a large number of antennas have a reasonable form factor while providing large beamforming gain and multi-user MIMO (MUMIMO) capability. However, the THz propagation mechanisms differ in various ways from previously explored channels. In fact, reflections by a rough surface and diffuse scattering mechanisms are the most critical features contributing to spatial and temporal dispersion at THz frequencies. For this study, first, we develop a hybrid modeling approach of 3D ray-tracing simulations in a realistic office room at 300 GHz and 350 GHz to obtain estimates of the impact of surface roughness on the channel capacity which depends on generic channel characteristics. Then, the indoor multipath propagation and its impact on massive MIMO channels considering smooth and rough surfaces are investigated by employing the Beckmann-Kirchhoff (B-K) model. Finally, channel capacities of indoor THz massive MIMO channels with different surface roughnesses for both line-of-sight (LoS) and non-line-of-sight (NLoS) scenarios are calculated. Though based on modeling results, the scattering properties of materials can favorably be exploited to maximize spatial multiplexing gains, especially in LoS scenarios.
12 June 2017 in IEEE Transactions on Vehicular Technology
Hybrid analog-digital beamforming is a well-known cost-efficient signal processing method for massive MIMO systems. In this paper, we propose a hybrid beamforming massive MIMO system based on Rotman lens analog beamforming with beam selection and digital beamforming. Rotman lens is a low-cost true-time-delay analog beamforming network supporting wide bandwidth signals, which is thus more attractive than the analog beamforming networks based on conventional high-cost phase shifters. To study our system based on Rotman lens, we first examine two potential array architectures, i.e., full-array and subarray architectures, concerning the RF design feasibility, insertion loss, and system scalability. Since the beam selection is required in our system, we propose two beam selection algorithms, i.e., greedy search-based method and branch- and bound-based method, aiming to optimize the hybrid beamforming system in terms of error-rate performance for both the full-array and subarray architectures. To validate our system in practice, the proposed system is also investigated and verified experimentally. In particular, we design and fabricate a sample of Rotman lens operating in the 5-GHz band, whose measured results agrees well with the computer simulation results. The measurement results are incorporated into the Monte-Carlo simulation with the proposed beam selection algorithms to study the error-rate performance. Our simulation results show that the hybrid beamforming system using the low-cost Rotman lens has performance comparable to that of the system using the high-cost phase shifters, and exhibits wideband capability and superior performance over the small-scale MIMO system under the same number of RF transceivers.