Massive MIMO - Why Massive MIMO?

 
I think the main reason for Massive MIMO for 5G is 'there is no other choice'. It is highly likely that we will use very high frequency (mm Wave)
signal in 5G.

High frequency mean => that the size of single antenna will be very small and => the aperture (the area for receiving energy) will be
very small => we need to use a large number of transmission antenna.

To overcome this small aperture on receiver side at high frequency, we need to use a large number of transmission antenna.
This would be the main reason, but once we adopt the Massive MIMO technology, we can enjoy some other advantages coming from using a large
array antenna that will be described later.
 
Now let's look more into the meaning of 'there is no other choice'. I will describe on this aspect based on WNCG Prof. Robert Heath on Millimeter Wave MIMO
Communication (I recommend you to look into the presentation on YouTube).
 
Let's assume a situation where we have one transmission antenna and one receiver antenna placed with distance R as illustrated below.

 
- If the transmission antenna transmit signal with the power of Ptx, what would be the received signal power?
If we assume the ideal condition, the received power can be represented as follows. This would be very familiar form that you have known from
high school physics, so called squared inverse rule.
The received power is decreased in proportion to the square of the distance from the transmission antenna.
 For example, if the distance gets two times farther away, the received power gets decreased by 4 times.
 

 
It sounds very simple. This ideal equation does not contain any parameter about frequency (wave length) or the gain of the receiver antenna. It
means the received signal power is not influenced by signal frequency or receiver antenna gain.
But we know this is not true from our common sense in radio communication. In real life, the received signal power IS affected by the frequency
(wave length) and receiver antenna gain.
If we improve the mathematical model to include the frequency (wave length) and receiver antenna gain, the model can be described as shown
below. According to this equation, the received power is in proportion to the square of the wavelength.
 For example, if we assume the antenna gain does not change, the frequency gets increased by 2 times (this mean that the
wavelength gets shorten by 2 times), the received power gets decreased by 4 times.

 
As I mentioned before, we will use much higher frequency (meaning much shorter wavelength) signal in 5G. It means the received power will be
much lower than in current communication system.
 For example, if we use 2 Ghz frequency in current communication and we will use 20 Ghz frequency in 5G, the wavelength in 20 Ghz is 10
times shorter than the wavelength of 2 Ghz. It means the received power at 20 Ghz will be 100 times lower than the received power at 2
Ghz.
In reality, the situation gets even more complicated because not only the receiver antenna gains but also transmission antenna gain plays role
as well. If we add the transmission antenna gain into the equation, it would become as shown below.

 
Now let's think of how we can overcome the drastic received power reduction at high frequency. In other words, the question is 'how we can make
Prx larger ?'. Mathematically it is simple. You can get larger Prx by setting the parameters as follows.
i) Increase Ptx (Transmitter Power)
ii) Decrease the distance between the transmitter and receiver antenna.
iii) Increase wavelength (use low frequency)
iv) increase receiver antenna gain
v) increase transmitter antenna gain.
In real life, can we use all of these options? The answer is 'No'. Let's look at each of these options one by one and think which one can be
applicable in real life.
 
Option i) can be doable to a certain degree, but we cannot increase the transmitter power as much as we want.
Option ii) cannot be the solution since we cannot change the distance as we like.
Option iii) cannot be the solution. Once we (standard organization and each Network Operator) decided to use a certain frequency, we have to
follow it. We cannot change as we like.
Option iv) and Option v) can be a doable solution. It may not be easy to increase the antenna gain, but at least nobody (standard organization,
Network Operator) does not prevent us from trying to increase the antenna gain.
 
Then how can we increase the antenna gain? We may increase the antenna gain by design (e.g, shape, material etc), but the amount of gain
improvement by design cannot be as large as to compensate the huge amount of the power reduction cause by the increased frequency.
 Almost the only way in this case would be to increase the number of antenna and it is the major motivation of using Massive MIMO as
illustrated below.
In addition to increasing the received power, Massive MIMO provides several other advantages as well.
 
According to "Massive MIMO for Next Generation Wireless Systems" ([5]), the potential (advantage) of Massive MIMO is described as
follows:
 Massive MIMO can increase the capacity 10 times or more and simultaneously, improve the radiated energy-efficiency in the order of 100 times.
 Massive MIMO can be built with inexpensive, low-power consumption
 Massive MIMO enables a significant reduction of latency on the air interface (due to robustness against fading)
 Massive MIMO simplifies the multiple-access layer
 Massive MIMO increases the robustness both to unintended man-made interference and to intentional jamming.
 
I found another very well summarized list of motivation and challenges about Massive MIMO from 3GPP R1-136362 ([6]).
 If we assume that we are using a fixed antenna size relative to the wavelength (e.g, size of 1/4 wavelength, 1/2 wavelength etc), as the carrier frequency
goes higher, the path loss increases.
 This means the absolute physical size of the antenna gets smaller as carrier frequency goes higher. It means we can put more antenna in the same
area in higher carrier frequencies. Based on these facts, we may compensate the high path loss in high carrier frequencies by putting more antenna
without increasing the total size of the antenna array.
 As the carrier frequency increases beyond roughly 10 Ghz, diffraction will no longer be a dominant propagation mechanism. In this frequency, reflection
and scattering will be the most important propagation mechanism for non-line-of-sight propagation link.
 As the carrier frequency goes higher, the penetration loss from propagating into a building tends to increase. This would make in-building coverage
impractical for BSs deployed outdoors.
 With massive MIMO (i.e, using many antennas in the array), we can implement high gain adaptive beamforming that would produce the effect of increasing
the coverage and create less interference in the system (because the beamwidth gets very narrower).

Spatial Focus with More Antenna

 
There is one thing that is automatically acquired by Massive MIMO. It is the fact that most of energy transmitted from the antenna array focus on
very narrow area. It means the beamwidth get narrower as you use more antenna.
Following plot would give you an example for the effect of beamwidht narrowing with the increased number of antennas.
This effect would cause both advantage and disadvantage at the same time. Advantage would be that there will be less interference between
beams for different users since each of the beam would be focused in very small area and the disadvantage would be that you have to implement
very sophisticated algorithm to find exact location of the user and directing the beam to the user with high accuracy.

(NOTE: Click on the picture for slideshow / animated version in my visual note: two difference examples are there. One for beam forming and the
other one for beam steering).

Note *: In this example, I assume that each of the antenna transmit the exact same power regardless of whether it is in 2 antenna array or in 4
antenna arrays. So, you see higher peak power in 4 antenna arrays. But in reality, they would decrease the transmit power for each antenna as
they increase the number of antennas. The point is that you should not increase the total transmitted power from the whole array even though you
increase the number of antennas.
 
Following is the matlab source code for this example. Don't take this seriously.. this is just a toy program to give you the big picture of how
antenna array works. it would not be accurate model. Just play with it and have some fun.
 
Save the following code in a m file named waveCosPh.m
function z = waveCosPh(x,y,Ph,r)
 d = sqrt(x.^2 + y.^2);
 
 dim = size(d);
 dr = dim(1);
 dc = dim(2);
 
 for i = 1 : dr
     for j = 1 : dc
         if d(i,j) > r
            d(i,j) = pi/2;
         end;    
     end;     
 end;    
 z = cos(d-Ph);
end
 
Run following code in another m file (any name is OK) that is located in the same folder as waveCosPh.m
clear all;
 
xstep = -10*pi:pi/10:10*pi;
ystep = -10*pi:pi/10:10*pi;;
[X,Y] = meshgrid(xstep,ystep);
 
[X1,Y1] = meshgrid(xstep-pi/2,ystep);
Z1 = waveCosPh(X1,Y1,0,10*pi);
 
[X2,Y2] = meshgrid(xstep+pi/2,ystep);
Z2 = waveCosPh(X2,Y2,0,10*pi);
 
[X3,Y3] = meshgrid(xstep-pi/2-pi,ystep);
Z3 = waveCosPh(X3,Y3,0,10*pi);
 
[X4,Y4] = meshgrid(xstep+pi/2+pi,ystep);
Z4 = waveCosPh(X4,Y4,0,10*pi);
 
Z = Z1 + Z2;
 
subplot(2,3,1);
mesh(X,Y,Z);xlim([-10*pi 10*pi]);ylim([-10*pi 10*pi]);zlim([-4,4]);
view([-40 70]);
 
subplot(2,3,2);
mesh(X,Y,Z);xlim([-10*pi 10*pi]);ylim([-10*pi 10*pi]);zlim([-4,4]);
view([-40 10]);
 
subplot(2,3,3);
mesh(X,Y,Z); xlim([-10*pi 10*pi]);ylim([-10*pi 10*pi]);zlim([-4,4]);
view([0 90]);
 
Z = Z1 + Z2 + Z3 + Z4;
 
subplot(2,3,4);
mesh(X,Y,Z);xlim([-10*pi 10*pi]);ylim([-10*pi 10*pi]);zlim([-4,4]);
view([-40 70]);
 
subplot(2,3,5);
mesh(X,Y,Z);xlim([-10*pi 10*pi]);ylim([-10*pi 10*pi]);zlim([-4,4]);
view([-40 10]);
 
subplot(2,3,6);
 mesh(X,Y,Z); xlim([-10*pi 10*pi]);ylim([-10*pi 10*pi]);zlim([-4,4]);
view([0 90]);
 
Following is another toy program that shows you beam pattern out of 2 dimensional antenna array (this is linear scale, not dB scale). You would
notice how narrower the beam width become as the number of antenna in the array get larger. (Note : I put more examples of various beams with
varying number of antenna configuration at my visual note. See here, here, here, here, here, here, here, here, here, here)
 

 
Followings are the matlab code that I used to draw the pattern shown above. You may play with the code by changing M, N value.
N = 16;
M = 16;
 
phi = linspace(-pi,pi,10*N);
theta = linspace(-pi,pi,10*N);
[phi,theta] = meshgrid(phi,theta);
 
 
Beam = sinc(1/pi*N*phi/2)./sinc(1/pi*phi/2).*sinc(1/pi*M*theta/2)./sinc(1/pi*theta/2);
Beam = abs(Beam)/max(max(abs(Beam)));
 
subplot(1,3,1);
[x,y,z] = sph2cart(phi,theta,Beam);
surface(x,y,z,'EdgeColor','none');axis([-1 1 -1 1 -1 1]);
xlabel('x');ylabel('y');zlabel('z');
grid();
view(-45,70);
 
subplot(1,3,2);
[x,y,z] = sph2cart(phi,theta,Beam);
surface(x,y,z,'EdgeColor','none');axis([-1 1 -1 1 -1 1]);
xlabel('x');ylabel('y');zlabel('z');
grid();
view(0,90);
 
 subplot(1,3,3);
[x,y,z] = sph2cart(phi,theta,Beam);
surface(x,y,z,'EdgeColor','none');axis([-1 1 -1 1 -1 1]);
xlabel('x');ylabel('y');zlabel('z');
grid();
view(0,0);
 
You can get more theoretical explanation about this property from various papers and following is one example based on  Scaling up MIMO :
Opportunities and Challenges with Very Large Arrays
 
Following is to illustrate how large antenna arrays can focus the electromagnetic field to a certain geographic point. This specific case shows the
resulting normalized field strength. (This example uses TRBF (Time-Reversal BeamForming) with MF precodings.
 

 
This result shows
i) the field strength can be focused to a point rather than in a certain direction
ii) more antennas improve the ability to focus energy to a certain point
With enough many antennas and favorable propagation
i) TRBF focus power and yield a high spectral efficiency through spatial multiplexing to many terminals
ii) TRBF also reduce (or completely eliminate in ideal case) inter-symbol interference, meaning that we could dispense with OFDM and its
redundant cyclic prefix.
 
Process being done by each base station antenna
i) convolve the data sequence intended for the k-th terminal with the conjugated, time-reversed version of his estimate for the channel
impulse response to the k-th terminal
ii) sum the K convolutions
iii) feed that sum into the antenna
 
 
 
Reference
 
[1]  GFDM Interference Cancellation for Flexible Cognitive Radio PHY Design
R. Datta, N. Michailow, M. Lentmaier and G. Fettweis
Vodafone Chair Mobile Communications Systems,
Dresden University of Technology,
01069 Dresden, Germany
Email: [rohit.datta, nicola.michailow, michael.lentmaier, fettweis]@ifn.et.tu-dresden.de
[2]  5G NOW. D3.1 5G Waveform Candidate Selection
 
[3]  Massive MIMO and Small Cells: Improving Energy Efficiency by Optimal Soft-Cell Coordination
Emil Bjornson, Marios Kountouris and Merouane Debbah
Alcatel-Lucent Chair on Flexible Radio, SUPELEC, Gif-sur-Yvette, France
Department of Telecommunications, SUPELEC, Gif-sur-Yvette, France
ACCESS Linnaeus Center, Signal Processing Lab, KTH Royal Institue of Technology, Stockholm, Sweden
 
[4]  Massive MIMO Info Point
 
[5]  Massive MIMO for Next Generation Wireless Systems
Erik G. Larson, ISY, Linkoping University, Sweden
Ove Edfors, Lund University, Sweden
Fredrik Tufvesson, Lund University, Sweden
Thomas L. Marzetta, Bell Labs, Alcatel-Lucent, USA
 
[6]  Scaling up MIMO: Opportunities and Challenges with Very Large Arrays
Fredrik Rusek, Dept. of Electrical and Information Technology, Lund University, Lund, Sweden
Daniel Persson, Dept. of Electrical Engineering (ISY), Linkoping University, Sweden
Buon Kiong Lau, Dept. of Electrical and Information Technology, Lund University, Lund, Sweden
Erik G. Larsson, Dept. of Electrical Engineering (ISY), Linkoping University, Sweden
Thomas L. Marzetta, Bell Laboratories, Alcatel-Lucent, Murray Hill, NJ
Ove Edfors, Dept. of Electrical and Information Technology, Lund University, Lund, Sweden
Fredrik Tufvesson, Dept. of Electrical and Information Technology, Lund University, Lund, Sweden
 
[7]  3GPP TSG-RAN WG1 #85   R1-165362 : Multi-Antenna Architectures and Implementation issues in NR
 
[8]  Massive Mimo Blog by EMIL BJÖRNSON

Further Studies
 
It seems certain that we will employ the Massive MIMO as one of the core technologies in 5G. However, it doesn't
mean that this technology is already mature (complete). There are many things to be improved or resolved about this
technology. This page would list up some of the area that are commonly listed as further study items.
 
 How to arrange Antenna? 
 
As you know, in Massive Antenna you would have huge number of antennas. Now you would have questions .. how
should I arrange those antennas to achieve the best performance?
 
Following illustration shows the various type of antenna arrangement that I have seen from various technical
materials.

 What would be the best arrangement? Will there be any new method of arrangement?
 
These are questions that should be answered from further research.
 
 
 How to model the 3D Channel?
 
If you arrange the antenna as (B), (C), (D), you would point the direction of the beam both in horizontal and in
vertical direction. If you combine the two directions, you can point the beam to any direction in 3D space (at least
almost of half of 3D sphere). It is good, but there is complication as well. Now you need to consider the channel
factors in all of those direction and you would need mathematical models to take those 3D factors into account.
This kind of channel models are one of the areas that require a lot of further study.

 How to apply it for FDD Operation?

 
I think this is the biggest drawback of Massive MIMO (at least as of now). In order to perform the best beamforming,
you need to have accurate (detailed) information of channels that is continuously changing. In order to get this kind
of information, you need to get the report from UE on downlink channel quality.

To do this, you need to allocate a lot of resources for downlink reference signal which would cause serious waste
of resource.
In FDD, we don't have any good idea to get the channel information without using this kind of channel
quality report based on reference signal.
However, in TDD, we can use some alternative technology which may not require this kind of UE reporting.
In TDD, we are using the same frequency band for both downlink and uplink.
- So, if Network can estimate the uplink channel quality from UE transmission signal, you can use that
information as downlink channel quality. Therefore, in TDD you can create pretty optimized beam without
getting explicit channel quality report from UE.
Of course, the estimation derived from uplink signal may not be exactly same as downlink signal because timeslot for
uplink and downlink is different. So, the channel estimation for UL at a certain timeslot may not be exactly
same as the downlink slot. However, this is the most commonly accepted and practiced idea as of now.
 
Due to this reason, most of Massive MIMO implementation are being done in TDD mode.
 
 
  How to generate wide beam from a large array?
 
One of the key ideas behind Massive MIMO is to increase Antenna gain by constructively adding up multiple antenna
output to a single beam and by this process the width of the resulting beam tends to get narrower. We can say this
narrow beam is good in terms of energy density, but it also means the area covered by a beam would be very narrow.
It means that the beamforming and directing should be very quick and accurate to properly focus on the target UE,
but this is not always simple and easy especially when the UE is in fast moving condition.
So, it would be necessary to widen the beam width without sacrificing too much of the performance of the
massive MIMO.
 
How to Calibrate the Antenna System?
 
Anybody who has experience of RF/mmWave design or testing would understand that the complexity and difficulties
of design/testing would increase exponentially as you have more signal path. Even assuming that the design is
properly done, you have to make it sure that all of the signal path and antenna are properly calibrated in order for the
antenna system to work as intended. Calibrating those huge number of antenna path is definitely challenging task.
 
How to handle the complexity of Scheduling and Precoding?
 
As you know, the biggest motivation of the Massive MIMO is to increase the directivity and gain for specified target
devices.
Another motivation (or requirement caused by beam forming) is to implement MU-MIMO (Multi-User MIMO).

However, the scheduling and Precoding would get more complicated as more antenna is used and more user is
targeted. How to handle this kind of situation would be a big question. Just to increase DSP power? or to come up
with a new/smart mathematical method to handle this without increasing DSP requirement too much?

What is MU-MIMO?
 
MU-MIMO stands for Multi User MIMO. It means Performing MIMO for more than 2 UEs simultaneously as illustrated
below. This is not a new concept. We have MU-MIMO in current LTE (TM5) and WLAN (802.11ad).
However, the scale of MU-MIMO will be much larger, and deployment will also be more common.
- As far as I experienced, I haven't seen any case where TM5 is really used in current LTE live network.
- In case of 802.11ad, the distance between UE and transmitter antenna is designed to be very short
comparing to 5G network.

So real implementation of MU-MIMO for 5G will be much more challenging.

 

 
How challenging it would be to implement MU-MIMO? The answer would be different depending on many factors. Even
with the same number of users and same number of Tx/Rx antenna, there can be different mode of antenna
allocation as illustrated below. There can be several factors to be considered in MU-MIMO implementation as follows.
 How many UE should be covered?
 How many Tx antenna and Rx antenna will be used?
 What kind of Receiver design (Equalizer design) will be used?
 What kind of Precoding Algorithm will be used?