Beneath the Tip of the Iceberg - What you should know before deploying LTE devices
We‘ve learned a lot from 3G rollouts. Some went smoothly and began to pay off very quickly. Many required a bit more work. Too often, early-adopter subscribers invested in faster technology but struggled with the basic service.
"When problems with a new service are discovered only after deployment, it takes significantly more time and money to recover."
LTE is a more complex and more hyped rollout than 3G was. So fixing the problems before subscribers find them is that much more critical to successful LTE business models.
Download the white paper to find out what you need to know before deploying LTE Devices
No Hidden Surprises
Beneath the Tip of the Iceberg
What you should know before deploying LTE nullenullices
1
Introduction
We‘ve learned a lot of lessons from
3G rollouts. Some went smoothly
and began to pay off very quickly.
Others required a bit of work
when early-adopter subscribers
invested in faster technology but
had problems getting basic service.
One thing we learned from the
overall deployment of 3G is that
when there is a problem on the day
of deployment it takes a lot of time
and money to recover. Obviously,
LTE is more complex than 3G, so the
lessons we’ve learned are even more
critical to successful deployment.
2
What nulloes LTE nullean nullor nullobile nullenullicesnull
The nullromises of LTE
While many of us take the use
of cellular data applications
for granted, we are in the
minority. Cellular data usage
is in its infancy. However,
the public’s fascination with
mobile data applications
has led to projections that
global cellular data traffic will
roughly double every year for
the next five years, resulting
in an astounding projected
usage of over 2 exabytes
(2,000,000,000,000,000,000
bytes) per month in the year
2015. How can we successfully,
effectively and efficiently
support the increasing
expectations of data users?
Long-Term Evolution (LTE)
technology promises to address
key areas of data delivery so that
network operators and device
manufacturers can deliver on the
promise of LTE and still make
a profit. As one example, LTE
promises tremendously efficient
data delivery; using a 20 MHz
MIMO-enabled channel, we hope
to see downlink data rates greater
than 100 Mbps, roughly an order
of magnitude better than we see
in many of our networks today.
LTE also offers shorter delay
or latency, which is important
since a lot of applications,
such as Voice over IP (VoIP)
and gaming will be real-time
applications. Finally, LTE
delivers a seamless mobile
experience even as the user
moves between regions of
LTE coverage and regions of
legacy-only coverage. But
for all of these objectives
or promises to be met, a lot
of testing challenges and
performance challenges still
need to be overcome.
3
Implications of nullThe nullomisesnull
to the nullobile nullenullice
One big difference between LTE and
3G technologies is that almost all
of LTE’s features and functionality
will be new and available on the
day of the technology launch.
3G rollouts have taught us that
the world long remembers what
happened on “launch day”, so
early failures are not acceptable.
This adds an additional burden;
the development of all relevant
network element (including UE’s)
is occurring in parallel and on
very aggressive schedules.
LTE devices are also much more
complex than any we’ve previously
seen. Devices which can support
GSM, R99, HSPA, HSPA+ must
now also support LTE; this also
means that the UE must support
additional frequency bands.
Yet another implication is the
fact that Multiple-Input, Multiple-
Output (MIMO) antenna technology,
currently thought of as a “nice-
to-have” feature, becomes a
mainstream requirement of the
UE, directly impacting the design
of antenna configurations. All
of this creates a very complex
scenario for an extremely intricate
device design, and all of it must be
working acceptably on launch day.
4
LTE nullenullices nullTimeline
In late 2009, TeliaSonera launched
the first LTE networks in Sweden
and Norway. In 2010 we’ll see other
significant LTE networks coming
online, but with data-only devices:
USB dongles and embedded modules,
such as those used in machine–to-
machine (M2M) applications.
In 2011 we expect to see the first
handsets and smart phones. They
will provide not only LTE IP data
services, but also some type of
voice capability, whether that
be voice-over-IP, circuit-switched
fall-back or simultaneous voice/
circuit-switched applications while
running IP services over LTE.
2012 is when we’ll see
most of the form factors
traditionally observed in wireless.
We may also see the introductions
of new LTE bands, new feature
sets and more permutations of the
way voice services can be provided
in the multi-mode LTE device.
2010
• Data only
• No voice services
• Mobile dongle
• PCMCIA cards for
Laptops, PC
• Embedded modules
2011
• Limited handset
only
• 2G/3G CS
Fallback only
2012
• Broad handset
deployments
• LTE advanced
• IMS based voice
solutions
• CS Fallback with
Intelligent Voice
Call Continuity (VCC)
5
The Impact of nullenullice Testing
Globally, the most familiar type of
device testing is GCF certification.
This is based on validated test
cases, a validated test process and
accredited test labs performing
minimum threshold testing under
specific conditions to verify the
conformance of the device to known
standards. As an example, the 3GPP
36.521-1 test specification is already
defined for conformance testing.
At the current time, the industry is
focused on the 3GPP’s December ‘09
core specification and the 3GPP is
releasing updates for compliance to
this spec. Based on this information
it is expected that validated testing
can begin sometime in the first
half of 2011, long after several
large-scale deployments have
been introduced to the public.
A second area of testing is referred
to as operator-certification or
acceptance testing. This is largely
driven by operators in North America
and in the Asia-Pacific region who
create their own test methodologies
to focus on deployment-specific
challenges. These test cases often
go above and beyond the minimum
performance criteria established
in the GCF certification process.
6
While the idea of operator
certification is less common
in Europe, some of the larger
European carriers show interest in
supplementing the GCF certification
process. One of the most
notable areas is in field testing.
Traditionally, after a device has
passed GCF certification, operators
perform live drive testing using an
active live network deployment.
Today, however, some operators
are interested in “virtual drive
testing”, where some of that
field testing is replicated in a lab
environment on test equipment.
This can reduce some of the
time, cost, and uncertainty
involved in field testing on live
infrastructure. Some of this
testing is already underway in
order to meet some commercial
launch-date targets for LTE UE’s.
The last category is UE performance
testing. One promise of LTE is the
ability to deliver advanced data
services. Both device manufacturers
and network operators see this as
the area in which to differentiate
offerings from their competitors’. For
a number of years Spirent has been
involved with third-party analysis of
UE performance, some of which
documented shocking differences
in the throughput capabilities
between commercial UE’s.
While the bulk of these studies
involved HSPA and HSPA+ devices,
the fact that LTE is more complex
leads to the expectation of an
even wider range of performance
differences. In any case, the moral
is clear; there is a tremendous
amount of variation in the
capabilities of certified devices,
and the public’s perception of a
device’s quality is directly tied
to performance in key areas.
7
nullotential nullitfalls in LTE nulleployment null nullobility
The public’s expectation is that an
LTE-capable device can be carried
off a plane, turned on anywhere
in the world and effectively,
efficiently, and quickly figure out
where it is, what type of service
it should acquire and will how
to register with the network.
System selection and re-selection,
then, is very important for these
devices which now support more
bands and modes than previous
devices. If a device starts off in a
3G region and then moves into an
area with LTE coverage, the device
must be able to detect the change
in order to provide the higher level
of service that LTE can offer.
This is a bit more complicated
than it sounds. This is not just the
addition of a single handover mode;
for a device that supports x radio-
access technologies, the number
of possible Inter-RAT transitions
is (x
2
-x)/2 . In other words, each
addition of a single new technology
creates an ever-increasing number
of Inter-RAT transition scenarios.
SOFT
HANDOVER
INTER-RAT
HANDOVER
INTER-RAT
HANDOVER
C
ell 1
C
ell 2
C
ell 3
C
ell 4
W
C
D
M
A W
C
D
M
A
W
C
D
M
A
l
t
e
8
Is the highest
priority system
3GPP or 3GPP2?
What systems are
available in my
current country?
Wake Up:
What country
am I in?
3GPP
Use
PLMN
MLPL Table
MSPL Table
Use
PRL
Country MSPL Index
MCC
1
1
MCC
2
2
MCC
3
3
Index System Type Priority Class
1 LTE Home
UMTS Preferred
CDMA Home
CDMA Preferred
Some important questions that
must be answered are: Can we
successfully establish those LTE
sessions? Does the device attempt
to transmit data services while in
a dormant or idle state? Can the
device successfully move from an
LTE network to a WCDMA data
network and back? What about
from one LTE network to another?
Can the device maintain data
services so transparently that the
end-user never even has to notice?
Can it properly prioritize choices
when multiple services are available?
When the device “wakes up”, does it
properly scan multiple technologies
for a country code? Can it properly
find the types of services (and
priorities) given this information?
Can the Quality of Service (QoS)
that is established on one network
be maintained during a transition
to another network? Does the
device obey the rules of roaming
agreements between operators?
Can connections be maintained
at the cell boundary? Should it
use PLMN-based or PRL-based
acquisition? And finally, can all
these decisions be made within
a reasonable time frame?
9
nullnull Implications in nullInullnull
Multiple-Input Multiple-Output
(MIMO) MIMO technology is not
new for LTE, but it is an extremely
significant factor in the success of
LTE. When LTE promises 100 Mbps
peak rates in the downlink, it largely
assumes that at least 2x2 downlink
MIMO is available and working
successfully. MIMO is also a radical
change to the fundamental Single-
Input Single-Output SISO technology
used in cellular up to this point.
This is a “game-changer” when it
comes to successfully testing device
performance. LTE UE’s are fairly
complex multimode devices. They’ll
physically be moved in seemingly
random, complex ways. Even static
devices (e.g., those used in machine-
to-machine [M2M] applications)
have to contend with constant
changes in the RF environment.
MIMO makes this both more critical
and more complex than ever.
Now the geometrical relationship
between antennas within a device
and antennas in the eNodeB adds
an additional complication that
previously hasn’t required testing.
If that’s not enough, all of this is
just a starting point. While our first
implementations of LTE will use 2x2
MIMO (two antennas on an eNodeB
and two on the device) plans for the
near future include 4x2 and possibly
even 4x4 implementations of MIMO.
z
z
x
y
x
10
This ultimately means that some
of the methods traditionally used
to test receiver performance are
no longer sufficient. Of course
we must account for baseband
performance of a device by using
a traditional conducted-signal test
scenario, but we now need to also
account for antenna performance
and antenna geometry.
One aspect of this testing is
enhanced fading emulation for the
test-bed. Since MIMO depends on
the spatial configuration of the RF
paths (essentially using physical
space as a domain in which to
multiplex data streams) channel
emulation must now account for
spatial channel characteristics.
In addition, as antennas move
relative to each other, angles
of arrival and departure change
quickly and unpredictably. It’s more
important than ever to dynamically
adapt models to emulate a time-
variant RF environment.
The certification process does
include MIMO test cases and it
does use channel models for
testing different MIMO modes. But
dynamic models are not a part of
the validation process, nor are some
of the spatial aspects of testing.
The way to enhance MIMO device
testing is to take the experience
of how devices operate in live
environments and use that to
enhance the models used in the
lab. This is done by adding dynamic
spatial channel models, real-world
complex MIMO correlation matrices
obtained in the field and brought
into the lab to enhance our testing.
11
nullInullnull nullnullernullThenullnullir nullnullTnullnull Testing
Traditional conducted-signal
testing bypasses antennas. This
may be sufficient for testing
single-antenna devices, but
LTE’s dependence on antenna
configurations and geometries
goes untested in these scenarios.
Testing in an over-the-air scenario
is critical when assessing the
combination of baseband receiver
and multiple antennas in the device.
The 3GPP is currently evaluating
several proposals for chamber-
based OTA testing, but the most
accurate test case seems to be one
which uses an anechoic chamber,
multiple transmitting antennas
and channel emulation to ensure
that the signals radiated from the
antennas are creating a realistic
signal field in the chamber.
There are other proposals currently
under consideration. For example,
one proposal is to use a “reverb
chamber” in which reflective
“paddles” are mechanically
rotated inside the test chamber.
While this approach appears to
be the cheapest to implement,
early studies show that it tends to
“average out” the effects of the
environment on MIMO, thereby
missing the “corner cases” that are
probably of the most interest.
12
nullirtual nullTnull
Yet another possibility is the concept
of virtual OTA, which takes OTA
testing one step further. In the
first step of Virtual OTA testing,
a device’s antenna configuration
is characterized by placing it in
an anechoic chamber fitted with
multiple transmitting antennas
(just as in standard OTA testing).
However, in Virtual OTA testing,
the device’s response to each
individual antenna is measured
and stored. This information
is then fed into an RF channel
emulator capable of replicating
the spatial channels that will seen
by the device upon deployment.
The result is an accurate emulation
of OTA RF channels so that realistic
testing can be performed in a
lab with conducted signals. The
goal is to achieve the benefits of
radiated or field testing, but with
the convenience of conducted
testing in a repeatable, reliable
lab environment. Virtual OTA also
promises to limit the amount of time
required with an anechoic chamber.
13
nullata
LTE is largely about providing
data services and providing
them well. Thanks to Spirent’s
work with industry analysts,
we’ve seen firsthand the wide
variation in the performance
of HSPA+ or HSPA devices.
Without global adoption of up-to-
date test methodologies, variations
in data performance from device
to device will be magnified as we
move to LTE. Remember, MIMO,
higher-order modulation schemes
and two-dimensional scheduling
are each intricate technologies
even in the best cases. The fact
is that upon deployment, each of
these features will be dynamically
adjusted (or will dynamically
adjust themselves) in real time
based on channel conditions
and feedback from the UE.
There is one more area of
data testing that can easily be
overlooked: the concept of “data
retry” testing. As data applications
continue to become both more
accessible and more popular with
mobile users, the danger to the
networks become more apparent.
There is currently a popular push to
eliminate the “walled gardens” of
cellular networks, where network
operators have tight control over the
applications they allow. Eventually it
is thought that operators may have
to open access to a large number
of data-centric applications over
which they have little control.
In addition, data rates are directly
related to the efficiency of MIMO
gains discussed previously. Testing
with legacy technologies also proves
that a device’s PC driver (for data
cards and modules) can have a
drastic effect, as can round-trip
delay times and (believe it or not)
the power source being used.
Finally, feedback from the UE to the
network can have an unexpectedly
significant impact. Studies
show that a single device which
“optimistically” reports channel
conditions will affect not only its
What happens, then, when a data
application tries to register with a
downed or unavailable server? Does
the device immediately begin to
hammer the network with access
requests which must be serviced?
What happens if the application
is a very popular one and many
types of devices begin generating
requests at the same time?
In networks where operators partner
with device makers, the solution is
to mandate limits on the frequency
with which the application can
request service. Data Retry testing
(sometimes called out as part of
Safe-for-Network test requirements)
nullata nullerformance
nullata nulletry
own data capabilities, but the data
rates of all the devices with which
it shares a cell. This has a significant
network-capacity cost to operators.
ensures that this feature is correctly
implemented across a wide variety
of potential network element
failures or access refusals.
Benchmarking analysis
indicates signinullcant
differences in the data
performance HSnullnullnull and
higher category HSnullnull
chipsetsnull
Michael Thelander, CEO,
Signals Research Group
14
Summary
LTE is a wholesale deployment of
new technologies at every layer,
usage case, and link in the network.
We can root out and avoid expensive
pitfalls by combining lessons learned
from 3G with our knowledge of the
fundamental new technologies.
1.
Mobility Testing – ensuring that the
device can correctly attach when
faced with multiple technologies
services, no matter where in the
world the mobile “wakes up”. This
testing also ensures that subscribers
can seamlessly and transparently
move between operator regions,
technology generations, and
even technology families (e.g.
from LTE to CDMA coverage).
2.
MIMO – RF and Access Network
considerations are often overlooked
by those charged with ensuring the
success of data transfer. However,
MIMO is a “game-changer”; it is
a new level of complexity at the
fundamental connection and it
is critical to LTE’s success.
3.
Data – Cellular subscribers read
“LTE” as “faster data and more
exciting applications”. Data
throughput must be tested in
realistic scenarios and in light of the
pitfalls outlined above. In addition,
Data Retry testing should ensure
that a rogue or failing network
element doesn’t lead to a cascade
of connection requests which can
bring the network to its knees.
The three key areas of interest arenull
These three areas are very much intertwined, but the situation is far from hopeless. Advances in test technology
and in testing philosophies offer the opportunity to make LTE successful from the day of deployment.
The areas of highest risk require
advanced testing beyond the
requirements of standards bodies.
The 3GPP does a great job of
ensuring that LTE will work, but
we (the industry) are responsible
for ensuring that it works well. We
are responsible for ensuring the
financially responsible deployments
of our products and services.
www.spirent.com
Americas
1-800-SPIRENT
+1-818-676-2683
sales@spirent.com
Europe and the
Middle East
+44 (0) 1293 767979
emeainfo@spirent.com
Asia and the Pacific
+86-10-8518-2539
salesasia@spirent.com
Spirent Communications
© 2010 Spirent Communications, Inc. All of the company names and/or brand names and/or product
names referred to in this document, in particular the name “Spirent” and its logo device, are either
registered trademarks or trademarks pending registration in accordance with relevant national laws.
All rights reserved. Specifications subject to change without notice. Rev. A 05/10
