Ralf R. Müller

Blind Pilot Decontamination

A subspace projection to improve channel estimation in massive multi-antenna systems is proposed and analyzed. Together with power-controlled hand-off, it can mitigate the pilot contamination problem without the need for coordination among cells. The proposed method is blind in the sense that it does not require pilot data to find the appropriate subspace. It is based on the theory of large random matrices that predicts that the eigenvalue spectra of large sample covariance matrices can asymptotically decompose into disjoint bulks as the matrix size grows large. Random matrix and free probability theory are utilized to predict under which system parameters such a bulk decomposition takes place. Simulation results are provided to confirm that the proposed method outperforms conventional linear channel estimation if bulk separation occurs.

Iterated Soft-Decision Interference Cancellation for CDMA

Multiuser detection by iterated soft-decision interference cancellation is known to achieve almost single user performance in low and moderate interference scenarios. In this paper, it is shown how this detection scheme can be further improved by adaptive calculation of the soft-decision rule. This enables a significantly higher number of active users signaling within the same bandwidth to be detected, implying that spectral efficiency is increased.

Low-complexity linear precoding for downlink large-scale MIMO systems

In this work, we present a low-complexity linear precoding scheme for downlink large-scale multiple-input multiple-output (MIMO) systems. The proposed scheme can achieve near minimum mean square error (MMSE) precoding performance in terms of the sum rate and is based on a matrix polynomial instead of matrix inversion. Simulation results show that matrix polynomials consisting of only a few terms are sufficient to closely approach the sum rate of the classical MMSE precoder and to perform orders of magnitude better than the simple conjugate beamforming (BF) precoder. We derive exact expressions for the computational complexity of the proposed scheme in terms of the number of additions and multiplications and compare it to the complexity of the BF and MMSE precoders. Our complexity analysis shows that for large number of base station antennas N compared to the number of generated transmit symbols τ per channel estimate and large number of users K, the proposed polynomial precoder has a lower complexity than the classical MMSE precoder.

Load modulated massive MIMO

We propose a new hardware architecture for cost-and size-effective implementation of massive MIMO transmitters based on load modulation that is fully compatible with standard receiver architectures. With load modulation, a massive MIMO transmitter can be driven by a single power amplifier operating at constant envelope. The various data streams are modulated onto the antennas elements by varying the complex impedances of the various antenna circuits at symbol rate. The law of large numbers ensures that the common power source is matched to the parallel concatenation of the various antenna circuits, if the number of antenna elements is very large. For 100 antenna elements that are supposed to transmit Gaussian-like signals, e.g. due to OFDM modulation and/or precoding, the crest factor is only 1.2 dB at a clipping probability of 0.1%. In that case, load modulation reduces the average consumed power and the amplifier peak power by 49% and 79%, respectively, as compared to classical distributed amplifiers with voltage modulation. Furthermore, one class F amplifier can be used to replace 100 class A/B amplifiers. Mutual antenna coupling need not be addressed by a physical matching network, but can be mitigated dynamically by digital signal processing in baseband without additional hardware. This allows for closer antenna spacing as in architectures with distributed amplifiers.

Robust Pilot Decontamination Based on Joint Angle and Power Domain Discrimination

We address the problem of noise and interference corrupted channel estimation in massive MIMO systems. Interference, which originates from pilot reuse (or contamination), can in principle be discriminated on the basis of the distributions of path angles and amplitudes. In this paper, we propose novel robust channel estimation algorithms exploiting path diversity in both angle and power domains, relying on a suitable combination of the spatial filtering and amplitude based projection. The proposed approaches are able to cope with a wide range of system and topology scenarios, including those where, unlike in previous works, interference channel may overlap with desired channels in terms of multipath angles of arrival or exceed them in terms of received power. In particular, we establish analytically the conditions under which the proposed channel estimator is fully decontaminated. Simulation results confirm the overall system gains when using the new methods.

Load modulated arrays: a low-complexity antenna

This article describes promising recent progress in the area of massive antenna array architectures with low front-end hardware complexity. The presented technology enables the design and implementation of antenna arrays with large numbers of elements, while obtaining significant front-end hardware savings as compared to the conventional solutions. This newly appearing design approach could be used in order to design either massive arrays with complexity that would be prohibitive with the current technology, or smaller arrays that offer high spatial degrees of freedom and are suitable for future small yet powerful cell nodes. RF hardware architectures with a single RF chain are reviewed, compared, and found superior to conventional MIMO implementations in terms of cost, dissipated heat, and physical size. The proposed improvements on the RF side allow the merging of the two dominant cellular technologies of virtual (distributed) and massive (centralized) MIMO into a hybrid approach of antenna arrays that is suitable for both large base stations and small (possibly cooperative) units such as remote radio heads.

Analysis of blind pilot decontamination

A nonlinear channel estimation scheme proposed by the authors in earlier work is shown to overcome the pilot contamination problem in massive multiple-input multiple-output (MIMO) systems. The method is based on a subspace projection using a singular value decomposition and is studied both by analytical and simulative means. The analysis presented in this paper is a refined version of an earlier work by the same authors. In addition, simulations based on a cellular network model show that the benefits of the proposed scheme are retained also in more realistic settings.

Power Randomization for Iterative Detection Over Random-Access Fading Channels

In this paper, we focus on throughput performance of multiuser communications systems over fading channels. More specifically, we consider the uplink where multiuser detection under asynchronous transmissions is exploited for random-access management and iterative receivers are considered for practical issues. Normalized throughput is evaluated through a semi-analytic procedure in order to avoid time-consuming simulations. Power randomization is explored as a means for improving performance through system asymmetries. It is found beneficial, particularly in case of overloaded systems operating at low-to-medium signal-to-noise ratio, where it allows reducing significantly the number of iterations at the receiver.

Low latency-constrained high rate coding: LDPC codes vs. convolutional codes

This paper complements and generalizes the results in [1], [2] while taking new formulas from [3] into account. We compare convolutional and LDPC codes of different rates with respect to their structural delay. This property is intrinsic to the encoding and decoding process and gives a fundamental limit to the achievable latency for end-to-end communication, even when infinite processing speed is assumed. We show that in the case of very low delay convolutional codes perform better than LDPC codes and even break fundamental existence bounds for block codes for some rates. Furthermore, non-punctured convolutional codes can benefit from their shorter constraint length and reduce the delay further.

RSB decoupling property of MAP estimators

The large-system decoupling property of a MAP estimator is studied when it estimates the i.i.d. vector x from the observation y = Ax + z with A being chosen from a wide range of matrix ensembles, and the noise vector z being i.i.d. and Gaussian. Using the replica method, we show that the marginal joint distribution of any two corresponding input and output symbols converges to a deterministic distribution which describes the input-output distribution of a single user system followed by a MAP estimator. Under the bRSB assumption, the single user system is a scalar channel with additive noise where the noise term is given by the sum of an independent Gaussian random variable and b correlated interference terms. As the bRSB assumption reduces to RS, the interference terms vanish which results in the formerly studied RS decoupling principle.