Paper Info
Title
When Are Low Resolution ADCs Energy Efficient in Massive MIMO?
Abstract
Massive MIMO (MaMI) is often promoted as a technology that will enable the use of low-quality, cheap hardware. One particular component that has been in the focus of MaMI-related research is the analog-to-digital converter (ADC), and use of very low-resolution ADCs has been proposed. However, studies about whether this strategy is justified from an energy-efficiency point of view have largely been inconclusive. In this paper, we choose system setup and models that re pect the hardware implementation reality as close as possible and perform a parametric analysis of uplink energy-efficiency as a function of ADC resolution. If antenna scaling and decrease of ADC resolution are considered independently, the energy efficiency is shown to be maximized at intermediate ADC resolutions, typically in the range of 4-8 bits. Moreover, optimal ADC resolution does not decrease when more antennas are used except in some specific cases, and when it does, the decrease is approximately logarithmic in the number of antennas. In the case when antenna scaling and ADC degradation are coupled through a constant-performance constraint, it is shown that energy efficiency cannot improve with reduced bit resolution unless the power consumption of blocks other than ADCs scales down with the upscaling of antennas at a fast enough rate. Altogether it is concluded that in MaMI, intermediate ADC resolutions are optimal in energy efficiency sense, and, except in some special cases, scaling up the antennas to very large numbers does not change this conclusion.
Year
DOI
Venue
2017
10.1109/ACCESS.2017.2731420
IEEE ACCESS
Keywords
Field
DocType
5G mobile communication,MIMO,energy efficiency,digital signal processing,circuits,analog-digital conversion
Flight dynamics (spacecraft),Computer science,Efficient energy use,MIMO,Effective number of bits,Electronic engineering,Logarithm,Automatic gain control,Electrical engineering,Scaling,Telecommunications link,Distributed computing
Journal
Volume
ISSN
Citations 
5
2169-3536
3
PageRank 
References 
Authors
0.37
18
3
Name
Order
Citations
PageRank
Muris Sarajlic162.15
Liang Liu29518.47
Ove Edfors32221134.61