Signal Processing - March 2016 - 21

Energy Efficiency (bits/J)

set to be the same for both sysbuild a massive antenna systems. The overhead of both
tem. And the other is to lever45
systems mainly comes from
age TR that inherently treats
TR System
40
Without
TR
the channel acquisition, and
the multipaths in the environ35
thus is similar. From Figure 5,
ment as virtual antennas.
30
it can be seen that, at the cost
Both can achieve the spatial25
of a larger bandwidth, the TR
temporal resonance at a par20
system can achieve comparaticular space and time that
15
ble if not better rates with the
we now commonly term as
10
genie-aided massive MIMO
the massive MIMO effect.
5
system by using only a single
Basically, it is a small focus0
antenna. This is achieved
ing ball of energy that takes
0
5
10
15
20
through exploiting a large
place due to the very high
SNR (dB)
number of virtual antennas that
degree of freedom.
naturally exist in the environTherefore, by exploiting a figure 6. A comparison of energy efficiency between a TR system and
ment. Note that the perforlarge number of virtual anten- direct transmission without TR.
mance of the TR system was
nas, a single-antenna TR sysobtained from real data, while that of
achievable rate between a practical TR
tem can achieve superior focusing effect
massive MIMO is the best case scenario.
system and an ideal genie-aided massive
in both time and spatial domains, resultAlso note that the massive MIMO sysMIMO system. The expected achievable
ing in similar promising performance as
tem requires a large number of antennas
rate is computed by averaging the achievmassive MIMO systems. In addition, the
that is suited for high-power outdoor base
able rate defined in [15] over different
implementation complexity of a TR sysstations, while the TR system leverages
channel realizations. By genie aided, we
tem is much lower since it utilizes the
large bandwidth to harvest naturally exmean an ideal condition that the interferenvironment as a virtual antenna array
isting multipaths, ideal for low-power inence and antenna coupling effects in the
and a computing resource. If cooperation
door applications.
massive MIMO system can be completeof users, e.g., cooperative communicaly eliminated with optimal beamforming.
tions, is a distributed way of achieving
The genie-aided massive MIMO system
the MIMO effect of high diversity, then
Energy efficiency
has M transmit antennas with 20-MHz
TR is similarly a cooperation of virtual
TR technology can take advantage of
bandwidth where M is in the order of
antennas to achieve the massive MIMO
the multipath propagation and achieve
hundreds [13], while the TR system has a
effect. The TR waveform is nothing but
good energy efficiency. The temporal
single transmit antenna with 1-GHz
to control each multipath (virtual antenfocusing effect concentrates a large porbandwidth. It is assumed that there are
na). Of course, what cooperation pays
tion of the useful signal energy of each
ten users in both systems, each equipped
for is the spectral efficiency due to the
symbol within a short time interval,
with a single antenna. In other words, the
use of time for distributed processing, in
which effectively reduces the ISI for
massive MIMO system we considered
return for the diversity effect.
high-speed broadband communications.
here is a multiuser MIMO (MU-MIMO)
In Figure 5, the performance comparThe spatial focusing effect allows the
system [2]. The total transmit power is
ison is shown in terms of the expected
signal energy to be harvested at the

1,600
LTE−A
LTE
TR System with
Basic Waveforms
TR System with
Optimal Waveforms

1,200

800

Achievable Rate (Mbits)

1,600

400

15
20
SNR (dB)
(a)

LTE−A
LTE
TR System with
Basic Waveforms
TR System with
Optimal Waveforms

1,200

800

400

15
20
SNR (dB)
(b)

figure 7. An achievable rate comparison: (a) one user case and (b) ten users case.
IEEE Signal Processing Magazine

March 2016

Table of Contents for the Digital Edition of Signal Processing - March 2016