MIMO and Beamforming in

the 5G Context
SBrT 2017
05/09/2017

Created by Will Sitch

Presented by Bruno Duarte
 A Brief History of Keysight

1939–1998: Hewlett-Packard years 1999–2013: Agilent Technologies years 2014+: Keysight years
A company founded on electronic Spun off from HP, Agilent became the On November 1, Keysight became an
measurement innovation World’s Premier Measurement independent company focused on the
Company electronic measurement industry

In September 2013, it announced the In April 2017, Keysight acquired Ixia to

spinoff of its electronic measurement strengthen its end-to-end software-driven
business testing and deliver insights into network
operations

Page
5G Drivers and Expectations
Massive
Growth in
For the User*
100x Data
Mobile Data
Demand
Rates
Amazingly fast
1000x
Massive
Growth in No. Great service in a crowd Capacity
of Connected
Devices Best experience follows you 100x
Densification
Exploding Super real-time and
Diversity of reliable communications 1ms Latency
Wireless
Applications
Ubiquitous things communicating Reliability
Dramatic *Courtesy of METIS 99.999%
Change in User
Expectations 100x Energy
of Network All founded on a solid
business model. Efficiency

Page
The importance of MIMO to deliver the 5G
It brings us more data
Shannon-Hartley Theorem:

Capacity = # Channels* BW * log2 (1 + S/N)

Increase data capacity by:

– Increasing #channels → MIMO

• Exploiting spatial multiplexing to deliver multiple streams of data within
the same resource block (time and frequency)
• Channel state information

– Increasing BW → mmWave frequencies require Beamforming

– Increasing S/N

Page
Types of Multiple Antenna Systems

SISO SIMO
- No diversity - Rx diversity
Tx Rx protection against - Rx smart antenna
Tx : Rx
fading (beamforming)
- Improved SINR

MISO MIMO
- Tx diversity - Tx/Rx diversity
- Tx smart antenna - Tx/Rx smart antenna
Tx : Rx (beamforming) Tx Rx
(beamforming)
: :
- Improved SINR - Spatial multiplexing
- Improved SINR
Or
Improved spectral
efficiency/data rates

© 2015 Keysight Technologies

Page
Single User MIMO
Multiple Spatial Channels for Higher Data Rates to 1 User
Note: This is a conceptual
implementation only. It
doesn’t take noise or non-
h00 square matrices into
s0 r0 ... s^0 account

h01
TX RX H-1
h10
s1 r1 ... s^1
h11

In this simple example, the

receiver is responsible for de-
R=HS
[ ]=[ ][ ]
r0 s0 multiplexing the two data
h00 h01
streams. The receiver does
or this with knowledge of the
r1 h10 h11 s1 ^ -1R
S=H channel [H]

© 2015 Keysight Technologies

Page
Multi User MIMO
Pre-coding Data for Multiple Users at the Same Time
Note: This is a conceptual
implementation only. It doesn’t
h00
s0 x0 r0 ... RX s^0 take noise or non-square
matrices into account
h01
W TX
h10 In this simple example, the
s1 x1 r1 ... s^ transmitter is responsible
RX 1 for pre-coding (W) the
h11 data using knowledge of
the channel [H]

^S= HX = HWS = (HW)S

[] [ ][ ] []=[ ][ ]
x0

x1
=
w00 w01

w10 w11
s0

s1
s^0

s^1
h00 h01

h10 h11
x0

x1
So it seems we want HW=I
(identity matrix) to allow for
non-square matrices, and
for inversion reasons use
pseudo inverse.
W = H-1 W = HT(HHT)-1

© 2015 Keysight Technologies

Page
MIMO Adoption and Evolution
– LTE MIMO has been in 3GPP standards since Release 8 (early 2009)

– LTE-Advanced (3GPP Rel. 10, 2011) supports 8 streams of DL MIMO

– LTE-A Pro (Release 13; early 2016) has Full Dimension MIMO (FD-
MIMO)

– Massive MIMO is a key component of 5G, currently being developed

Page
MIMO visualized Spatial Multiplexing:
Tx Diversity:
Different data, different paths
Same data, different paths

RRH

Page
Understanding Massive MIMO
Description: MU MIMO with Number of BS antennas >> number of UE’s
Motivation: Higher reliability, higher throughput, lower TX power,
simple single antenna UE design
The Graphics The Math
Overly simplified! Not completely intuitive!

Equations from “Noncooperative Cellular Wireless with Unlimited Numbers of

Base Station Antennas”, by Thomas L. Marzetta

© 2015 Keysight Technologies

Page
2D Massive MIMO, Free-space Path Loss Only
Reference Configuration with 4 Users: Total TX Power 0 dB Relative
15,000 λ

UE2
Victim UEs (hollow)
Target UE (solid)

2000 λ
50 omni elements
Linear Array
½ λ Spacing

Keysight Keysight

UE3

UE4
Keysight Keysight

© 2015 Keysight Technologies

Page
Massive MIMO Free Space
200 Ant,1/2 λ, Total Power Relative to Reference: -9.5 dB

UE2
UE1

Keysight Keysight

UE3

UE4

Keysight Keysight

© 2015 Keysight Technologies

Page
Observations of Massive MIMO
Number of Base Station Antennas

– Initial Analysis of 1, 2, 4, 8 an 16
base station antennas showed that
more antennas always improved
performance

– In the limit of an infinite number of

base stations, more antennas
continue to improve performance

– Studies trying to quantify the optimal

number of base station antennas
suggest a range of a “several
hundred” antennas may be the
optimal number

Source:
– HOW MUCH TRAINING IS REQUIRED FOR MULTIUSER MIMO? Thomas L. Marzetta
– Noncooperative Cellular Wireless with Unlimited Numbers of Base Station Antennas Thomas L. Marzetta
– Massive MIMO in the UL/DL of Cellular Networks: How Many Antennas Do We Need? Jakob Hoydis, Stephan ten
Brink, M´erouane Debbah

© 2015 Keysight Technologies

Page
Observations of Massive MIMO
Number of Base Station Antennas

– Initial Analysis of 1, 2, 4, 8 an 16
base station antennas showed that
more antennas always improved
performance

– In the limit of an infinite number of

base stations, more antennas
continue to improve performance

– Studies trying to quantify the optimal

number of base station antennas
suggest a range of a “several
hundred” antennas may be the
optimal number

Source:
– HOW MUCH TRAINING IS REQUIRED FOR MULTIUSER MIMO? Thomas L. Marzetta
– Noncooperative Cellular Wireless with Unlimited Numbers of Base Station Antennas Thomas L. Marzetta
– Massive MIMO in the UL/DL of Cellular Networks: How Many Antennas Do We Need? Jakob Hoydis, Stephan ten
Brink, M´erouane Debbah

© 2015 Keysight Technologies

Page
 Thank You !!!
Questions and Answers
Bruno Duarte
bruno.duarte@keysight.com

© 2015 Keysight Technologies

Page
Copyright © 2014 Agilent. All rights reserved
Observations of Massive MIMO
TDD vs. FDD

– In TDD, the training time is proportional to the number of user

terminals

– In FDD, the training time is proportional to the number of users plus

the number of base station antennas
• Both downlink and uplink channels need to be trained
• Training on the downlink channel scales with the number of base
station antennas

– The use of massive MIMO in FDD systems remains as one of the

key research areas currently under investigation

© 2015 Keysight Technologies

24 Page