Robust Location-Aided Beam Alignment in Millimeter Wave Massive MIMO

Abstract

Among the enabling technologies for 5G wireless networks, millimeter wave (mmWave) communication offers the chance to deal with the bandwidth shortage affecting wireless carriers. Radio signals propagating in the mmWave band experience considerable path loss, leading to poor link budgets. As a consequence, large directive gains are needed in order to communicate and therefore, beam alignment stages have to be considered during the initial phases of the communication. While beam alignment is considered essential to the performance of such systems, it is also a costly operation in terms of latency and resources in the massive MIMO (mMIMO) regime due to the large number of beam combinations to be tested. Therefore, it is desirable to identify methods that allow to optimally trade-off overhead for performance. Location-aided beam training has been proposed recently as a possible solution to this problem, exploiting long-term spatial information so as to focus the beam search on particular areas, thus reducing overhead. However, due to mobility and other imperfections in the estimation process, the spatial information obtained at the base station (BS) and the user (UE) is likely to be noisy, degrading beam alignment performance. In this paper, we introduce a robust beam alignment framework in order to exhibit resilience with respect to this problem. We first recast beam alignment as a decentralized coordination problem where BS and UE seek coordination on the basis of correlated yet individual measurements. We formulate the optimum beam alignment solution as the solution of a Bayesian team decision problem. We then propose a suite of algorithms to approach optimal designs with reduced complexity. The effectiveness of the robust beam alignment procedure, compared with classical designs, is then verified on simulation settings with varying location information accuracies.

Off-Grid Aware Channel and Covariance Estimation in mmWave Networks

“…2) Uniform Sampling of the Virtual Domain: In this approach, the AoA/AoD are taken from a non-uniformly quantized grid such that the (cos(θ rx ), cos(θ tx )) space appearing in the array response is uniformly quantized [10], [39]. The authors in [10] show that such grids reduce the coherence of the redundant dictionary due to preserving orthogonality which does not hold in the former approach.…”

Location-Aware Coordinated Beam Alignment in mmWave Communication

“…where the superscript BS and UE are used to indicate the observation at the BS and UE respectively. In this paper, we adopt a uniform bounded error model for the location estimation error [23], [28]. We assume that all the estimates lie within a disk centered on the estimated location.…”

“…However, for a mobile device, it is likely that the location information is noisy and designing beam alignment algorithms might not be straightforward as in [22]. A noisy location information is considered in [23], where the authors focused on an independent beam pre-selection at the BS and the UE. While the pre-selection algorithm can improve the beam selection speed, the performance of the selected beams cannot be guaranteed since the beam selection decision is weighed on the noisy location information.…”

Pilot contamination mitigation based on antenna subset transmission for mmWave massive MIMO

“…This assumption is related to beam alignment problem, the reader may see Maschietti et al 16 and Hashemi et al 17 for more details about this issue. This assumption is related to beam alignment problem, the reader may see Maschietti et al 16 and Hashemi et al 17 for more details about this issue.…”

“…For simplicity, the receiver and transmitter beams are assumed to be aligned for maximum reception, ie, w = a r ( ). This assumption is related to beam alignment problem, the reader may see Maschietti et al 16 and Hashemi et al 17 for more details about this issue. Replacing w in (3), we get the following:…”