Resurrection of 5G: In defense of Massive MIMO

CTN Issue: January 2016

Alan Gatherer, Editor in Chief, ComSoc Technology News

After our doom laden "death of 5G" series we did get some letters. In particular, the good people of the MAMMOET project in Europe managed to hit several points at once. So to start 2016 on an upbeat (and goodness knows it needs one), we have a positive look at massive MIMO covering both the throughput and the implementation for sub 6GHz. We hope this makes you feel just a little bit better about wireless in 2016. Comments (even doom laden ones) continue to be welcome of course.

the MAMMOET team with sub6G MIMO array

Massive MIMO clarified

A viable and profitable 5G wireless technology has to fulfill many requirements: high data rates per user, uniformly good service throughout the coverage area, ability to function under high mobility conditions, resilience to blockage and obstructions, and effective building penetration. It has become evident that known techniques such as the deployment of small cells and exploitation of higher frequencies are unable alone to address these needs. One technology that paves the way for 5G is Massive MIMO (MaMi), a new and most promising direction in mobile access [1].

This article aims to clarify the capabilities of MaMi. Technical proof of this is being developed in the research community, including the MAMMOET project [www.mammoet-project.eu], which is in the process of demonstrating that MaMi, operating in the below 6 GHz spectrum, without large new spectral resources and with inexpensive hardware, represents the ideal 5G technology.

The capacity and service offer: outstanding
Massive MIMO is the only 5G technology that can offer uniform quality-of-service in dense urban, suburban, and rural macro cells and that can serve mobile terminals which are occasionally affected by deep shadow fading or located at cell borders. It makes a clean break with current technology by using arrays with hundreds of antennas at the base stations, which simultaneously serve many tens of low-complexity terminals in the same time frequency resource through closed-loop spatial multiplexing/de-multiplexing (multi-user MIMO precoding/decoding).  Channel estimates are formed on the uplink through time-division duplexing (TDD) operation and reliance on reciprocity of propagation, which makes the operation scalable with respect to the number of antennas and leaves the channel coherence time as the only limiting factor. A very real impairment in multi-cell deployments is pilot contamination, which fortunately can be pushed to practical nonexistence in almost all scenarios of practical interest in mobile access applications [2]. An array power gain that arises by virtue of the MIMO precoding, improves the link budget to terminals having disadvantaged propagation conditions by tens of dB. Different base stations need not cooperate, other than for traditional resource allocation tasks. Conceptually, a 10x or more increase in gross throughput can be achieved with MaMi. Perhaps even more important is the significant gain in reliability due to flattening out of deep fades, hardening of the channel, and array gain [6]. This especially benefits cell edge users.

MaMi is not mmWave MIMO, even though mmWave systems definitely benefit from a high-gain antenna, or phased array. In contrast, MaMi may have its greatest impact in frequency bands below 6 GHz, where path loss and propagation are well understood and favorable, and hardware is mature and inexpensive. Spectrum below 6 GHz will always be the most valuable. In the United States, the spectrum auction for 65 MHz additional sub-2.2 GHz spectrum conducted by the FCC in 2015 totaled more than 40 billion USD. The market value of such spectrum will increase dramatically when the MaMi technology is introduced, since it provides a ten-fold and higher system throughput for a given spectrum.

Complexity and energy efficiency: lean
An obvious concern is how the large number of antennas (and associated transceivers and signal processing) will affect the complexity and energy consumption of the base station. In-depth analysis confirms the spectacular complexity and power reduction promise. This stunning improvement results from the fact that on the one hand, much less transmitted power is needed thanks to the array gain, and on the other hand, relatively low complexity hardware can suffice. For example consider the power amplifiers and their supporting circuitry.  In traditional base-station they are dominant consumers at about 1kW for 3 sectors because of the combination of the high output power need and the linearity requirements driving towards considerable back-off [3]. In MaMi a similar range can be covered with a total power consumption of 15W for 100 antennas.

Analog and radiofrequency (RF) components are expected to dominate MaMi energy consumption, which in absolute terms will decrease greatly. The system can operate with significantly lower overall transmitted RF power.  Also, the many constituent signals do not need to have high accuracy, and ADC and DAC resolutions as low as 3-4 bits are adequate. ADCs for these specifications can be realized with power consumption below 1 mW [4], which makes their total power negligible even if they come in hundreds. The full analog chains accordingly require only inexpensive low-power hardware. We anticipate that one can be realized with < 300mW [5], summing up to < 30W for the entire system.

Central digital signal processing benefits hugely, in terms of savings in hardware and power, from the simplicity of linear precoding/de-coding which perform nearly as well as costly algorithms such as successive interference cancellation. Avoiding these specifically in broadband systems results in significant hardware and power savings. Many of the per-antenna functions can benefit from optimized parallel processing and short word lengths.
Concluding, we are confident that the overall complexity and energy consumption in terms of J/bit can be lowered by a factor of 20 to 50 with respect to current macro base stations for the same range and capacity.

One of the great benefits of MaMi is that most of the complexity resides at the base station. MaMi terminals can be simpler than 4G terminals, and they achieve huge performance improvements using only a single antenna while operating with standard modulation techniques. The major difference lies in how the terminals experience the propagation channel. Compared to terminals in a traditional system, they will see large array gains and significantly reduced small-scale fading, owing to so-called channel hardening. The terminals can be made energy efficient, thanks to less complex receiver chains.

A 5G technology whose time has come
The throughput and energy efficiency benefits of Massive MIMO are unprecedented and the progress towards practical implementations is accelerating. Much remains to be done before operational systems emerge. Solutions are needed, for example, to efficiently cope with real traffic patterns and to optimally schedule users in real time. Importantly, the entire industry from hardware vendors to network operators needs to be fully engaged to empower the actual deployment. Fortunately, main industrial players have recognized the need for MaMi as a fundamental 5G technology component, and are taking this technology on to the standardization agenda, at first in the 3GPP context.

Now is a good time to discover this fascinating technology. We welcome your visit to www.massivemimo.eu to check or contribute to the MaMi literature and at www.mammoet-project.eu to follow up the progress towards practical implementations.

References:

  1. E. G. Larsson, F. Tufvesson, O. Edfors, and T. L. Marzetta, “Massive MIMO for Next Generation Wireless Systems”, IEEE Commun. Mag., vol. 52, no. 2, pp. 186-195, Feb. 2014.
  2. Xudong Zhu, Zhaocheng Wang, Linglong Dai, and Chen Qian, “Smart pilot assignment for Massive MIMO”, IEEE Communications Letters, Vol. 19, No. 9, September 2015, pp.1644-1647
  3. G. Auer, V. Giannini, C. Desset, I. Godor, P. Skillermark, M. Olsson, M. A. Imran, D. Sabella, M. J. Gonzalez, O. Blume, A. Fehske, “How much energy is needed to run a wireless network?”, IEEE Wireless Communications, vol. 18, no. 5, October 2011, pp. 40-49
  4. B. Verbruggen et Al., “A 60 dB SNDR 35 MS/s SAR ADC With Comparator-Noise-Based Stochastic Residue Estimation” The ADC achieves a 60.9 dB SNDR for a near-Nyquist input at 35 MS/s for a purely dynamic power consumption of 12 μW/MHz., IEEE Journal Of Solid-State Circuits, Vol. 50, No. 9, September 2015, pp. 2002-2011
  5. C. Desset, B. Debaillie, F. Louagie, “Modeling the hardware power consumption of large scale antenna systems”, 2014 IEEE Online Conference on Green Communications
  6. Emil Björnson, Erik G. Larsson, Mérouane Debbah, "Massive MIMO for Maximal Spectral Efficiency: How Many Users and Pilots Should Be Allocated?,"  IEEE Transactions on Wireless Communications, 2016.

On behalf of the MAMMOET project partners:
Liesbet Van der Perre, Sofie Pollin and Wim Dehaene (KU Leuven)
Erik G. Larsson and Emil Björnson (Linköping University)
Ove Edfors, Liang Liu, and Fredrik Tufvesson (Lund University)
André Bourdoux and Claude Desset (imec)
Franz Dielacher (Infineon), Javier Lorca Hernando (Telefonica), Klaus-Michael Koch (Technikon)

Scientific advisors:
Thomas L. Marzetta (Bell Labs, Alcatel-Lucent)
Piet Demeester (Ghent University - iMinds)
Jan Rabaey (University of California, Berkeley)

Editor-in-Chief: Alan Gatherer (alan.gatherer@huawei.com)

Comments welcome!

Leave a comment

Statements and opinions given in a work published by the IEEE or the IEEE Communications Society are the expressions of the author(s). Responsibility for the content of published articles rests upon the authors(s), not IEEE nor the IEEE Communications Society.