Project Details
Projekt
Compressive Covariance Estimation for Massive MIMO (CoCoMiMo)
Applicants
Professor Giuseppe Caire, Ph.D., since 8/2019; Professor Dr. Sjoerd Dirksen; Professor Dr. Holger Rauhut
Subject Area
Electronic Semiconductors, Components and Circuits, Integrated Systems, Sensor Technology, Theoretical Electrical Engineering
Mathematics
Mathematics
Term
from 2018 to 2023
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 404410006
Full duplex (FD) and massive multiple input multiple output (MIMO) are two technologies that are promising candidates for next generation (5G) cellular networks. In massive MIMO, a large number of antennas are employed at the base station (BS), which provides a high degree of spatial freedom, there by improving spectral efficiency. In FD systems, the nodes are allowed to transmit and receive in the same frequency-time channels, leading equally to higher spectral efficiency. Although FD massive MIMO systems promise a more significant improvement in the spectral efficiency, they are also faced with design challenges, in particular self-interference cancellation for FD systems and high-dimensional channel state information (CSI) acquisition for massive MIMO systems. An efficient precoder design lies at the heart of FD massive MIMO systems to exploit the spatial degree of freedom and to mitigate self-interference. We propose to design a precoder which uses the signal subspace of the users to reduce the complexity of the full exact CSI acquisition. Since the covariance matrix of the channel is approximately low rank, as the number of scatterers is usually small compared to the number of antennas in an FD massive MIMO BS, the signal subspace estimation problem can be considered as an estimation of a low-rank Toeplitz (1D antenna array) or block Toeplitz (2D antenna array) covariance matrix from subsampled and noisy training observations. We investigate theoretical guarantees for low rank (block) Toeplitz covariance estimation from compressed and noisy measurements. We will derive rigorous bounds on the estimation error in terms of the number of observed time samples and the number of sampled entries (antennas), with the prospect of inferring design guidelines for massive MIMO system parameters. This analysis is extended further to covariance estimation of sparse (block) Toeplitz matrices. We also address the power consumption and cost problem of high resolution Analog-to-Digital Converters (ADCs) by using one-bit ADCs and consider subspace estimation and precoding design based on one-bit quantized measurements/observations. For this purpose, we rigorously investigate low-rank (block) Toeplitz covariance estimation from (compressed, noisy) one-bit quantized samples from both experimental and theoretical perspective.
DFG Programme
Priority Programmes
International Connection
Netherlands
Ehemaliger Antragsteller
Professor Dr. Rudolf Mathar, until 7/2019
