Professional Documents
Culture Documents
HSPA/HSPA+
OPTIMIZATION
UMTS/HSPA+ CURRICULUM
This eBook is restricted to the use of a single authorized Vodafone staff member.
Do not distribute in electronic or print form.
VFN_745p
2100 Lakeside Blvd., Suite 300
Richardson, TX 75082 USA
Phone: +1.972.664.0727
Website: www.awardsolutions.com
If you have any questions, concerns or comments regarding this course please write to us at:
friends@awardsolutions.com
This course book and the material and information contained in it are owned by Award Solutions, Inc. (“Award Solu-
tions”) and Award Solutions reserves for itself and successors and assigns all right, title and interest in and to the
Award Content, Award Solutions’ logos and other trademarks, including all copyrights, authorship rights, moral
rights, publication and distribution rights, trademarks and other intellectual property rights. Award grants no license
or other rights in the contents of the course book or course, except as may be expressly set forth in a duly executed
written agreement between Award Solutions and the authorized user of this course book or the user’s employee or
principal. This course book shall not be modified, reproduced, disseminated, or transmitted by or in any medium,
form or means, electronic or mechanical, including photocopying, recording or any information retrieval system, in
whole or in part, without Award Solutions, Inc.’s express, prior written consent signed by an authorized officer
whose authority is evidenced by a duly signed corporate resolution.
This course book was designed for use as a student guide with the subject matter course taught by Award Solu-
tions’ authorized employees and contractors. It was not designed to be a standalone textbook. Award Solutions
makes no representations or warranties and disclaims all implied warranties with respect to the information con-
tained herein or products derived from use of such information and Award Solutions undertakes no obligation to
update or otherwise modify the information or to notify the purchaser or any user of any update or obsolescence.
To the extent permitted by applicable law, Award’s total liability in connection with the course and/or course materi-
al is the amount actually received by Award from the purchaser/user for the purchase or license of the course and
course material. This course book is not made for publication or distribution in the public domain and shall not be
published or placed in the public domain, in whole or in part, without Award Solutions, Inc.’s express, prior written
consent signed by an authorized officer whose authority is evidenced by a duly signed corporate resolution.
The 3GPP, LTE and LTE-Advanced logos are the property of Third Generation Partnership Project (3GPP). The
3GPP2 logo is property of Third Generation Partnership Project (3GPP2) and its organization partners. The
content of this document is based on 3GPP/LTE and 3GPP2 specifications which are available at
www.3gpp.org, and www.3gpp2.org.
About Award Solutions
AWARD SOLUTIONS, INC. has over 16 years of We offer a multitude of delivery methods and services:
training excellence in advanced wireless, IP, and network
transformation technologies. Our products and services ON-SITE TRAINING: Our Subject Matter Experts
provide our customers with innovative, flexible, and cost- travel to your facility to engage the students in
effective solutions that help rapidly boost workforce an interactive learning experience. This is a great
productivity to more quickly meet market demands. option for teams with 12 or more people.
Whether you are a training manager responsible for a large OUR PROMISE
organization, or a team lead responsible for enhancing your To continually demonstrate our core values: Integrity,
team’s skills, Award Solutions can meet your technology Expertise, Flexibility, Teamwork and Excellent Return on
training needs. Investment.
Network Transformation
4G LTE * Big Data Technology Overview........................................... 1 day
LTE Essentials..................................................................... 1 day Exploring Cloud Computing Service Models....................2 days
* LTE-Advanced Essentials.................................................... 1 day * Exploring Software-Defined Networking (SDN) and OpenFlow...2 days
* LTE RAN Performance Essentials....................................... 1 day Exploring the Service Oriented Architecture (SOA)..........2 days
* VoLTE Essentials................................................................. 1 day Exploring Infrastructure as a Service (IaaS).....................2 days
* LTE and VoLTE Essentials for Cell Techs...........................2 days * Exploring Network Functions Virtualization (NFV)............2 days
LTE Technology Overview...................................................2 days
LTE-Advanced Technical Overview....................................2 days
Exploring IPv6 for LTE Networks........................................2 days
VoLTE and IMS in LTE-EPC Networks................................3 days
UMTS/HSPA+
* Policy and Charging Control in LTE Networks...................2 days Exploring UMTS (WCDMA).................................................2 days
* Exploring IMS/VoLTE Networks.........................................2 days Mastering UMTS Radio Protocols and Signaling...................4 days
Mastering LTE Air Interface...............................................2 days
Mastering TD-LTE Air Interface..........................................2 days Mastering HSPA Protocols and Signaling.........................3 days
LTE Protocols and Signaling..............................................3 days HSPA+ Protocols and Signaling (R7, R8 & R9)..................2 days
LTE and GSM/UMTS Interworking.....................................2 days UMTS/HSPA/HSPA+ Air Interface.....................................3 days
LTE-EPC Networks and Signaling......................................3 days UMTS/HSPA+ RF Optimization Workshop........................4 days
* VoLTE Signaling and Operations: RAN/UE Perspective.......... 4 days
LTE RF Planning and Design Certification Workshop......5 days
LTE-EPC Capacity Planning Certification Workshop.........4 days
LTE RAN Capacity Planning Certification Workshop........3 days Wireless Landscape
LTE RAN Signaling and Operations Certification..............5 days Exploring Wireless Landscape, IP Convergence, and 4G....2 days
* LTE RF Optimization Certification Workshop (UE Based)...5 days Exploring Wireless Technologies and Networks..................5 days
* Small Cell and VoLTE RF Planning and Design Certification Workshop......4 days Fundamentals of RF Engineering......................................2 days
Exploring GSM/EGPRS/UMTS/HSPA/HSPA+...................5 days
Wireless and 3G/4G Basics............................................... 1 day
Emerging Trends
Wi-Fi Technical Overview...................................................2 days
* Small Cell Essentials........................................................... 1 day
* Small Cell Technical Overview...........................................2 days
* DAS Technical Overview.....................................................2 days
* New Course
4G LTE UMTS/HSPA+
Welcome to LTE (e).............................................................1 hour Welcome to UMTS (e).................................................. 1.5 hours
LTE Overview (e)...............................................................3 hours Overview of UMTS (e).......................................................2 hours
LTE SAE Evolved Packet Core (EPC) Overview (e)..........3 hours UMTS/WCDMA Air Interface Fundamentals (e).............3 hours
LTE Air Interface Signaling Overview (e).........................3 hours UMTS Signaling (e)...........................................................1 hours
VoLTE Overview................................................................3 hours UMTS Mobility (e).............................................................1 hours
Overview of IPv6 for LTE Networks..................................3 hours HSDPA (R5) (e).................................................................3 hours
HSUPA (R6) (e)............................................................. 2.5 hours
HSPA+ Overview (R7) (e).................................................4 hours
Emerging Trends
Overview of OFDM (e)......................................................2 hours
Multiple Antenna Techniques (e)....................................3 hours Wireless Landscape
Wi-Fi Overview (e).............................................................3 hours Welcome to Wireless Networks (e)...................................1 hour
* Small Cell Overview (e)......................................................1 hour Welcome to GSM/GPRS (e)......................................... 1.5 hours
1xEV-DO Networks (Rev 0) (e).........................................3 hours
1xEV-DO Networks (Rev A) (e).........................................3 hours
Overview of 3G Wireless Networks (e)........................ 1.5 hours
IP Convergence & IMS Overview of WiMAX (e).....................................................3 hours
Welcome to IP Networking (e).........................................3 hours
IP Convergence Overview (e)...........................................4 hours
Overview of MPLS (e)................................................... 3.5 hours
Overview of IMS (e)...................................................... 2.5 hours
Voice and Video over IP (VoIP) Overview (e)...................3 hours
IP Quality of Service (QoS) (e).........................................3 hours
Session Initiation Protocol (SIP) (e).................................2 hours
Ethernet Backhaul Overview (e)......................................3 hours
IP Basics (e)........................................................................1 hour
IP Routing (e)......................................................................1 hour
QoS in IP Networks (e).......................................................1 hour
TCP and Transport Layer Protocols (e)..............................1 hour
Ethernet Basics (e).............................................................1 hour
Ethernet VLANs (e).............................................................1 hour
Ethernet Bridging (e)..........................................................1 hour
Interconnecting IP Networks (e)........................................1 hour
Welcome to IPv6 (e)...........................................................1 hour
Chapter 1
Chapter 2
RF Measurements ......................................................................................................................................... 13
Chapter 3
Additional Charts............................................................................................................................................ 59
Chapter 4
Additional Charts............................................................................................................................................ 83
Chapter 5
Chapter 6
Chapter 7
ii
Table of Contents
Acronyms .....................................................................................................................................................187
References...................................................................................................................................................193
iii
1 | WCDMA Performance Optimization
Chapter 1:
WCDMA Performance
Optimization
1
1 | WCDMA Performance Optimization
Objectives
After completing this module, you will be able to:
• Explain the goals of optimization
• Sketch the basic optimization process
• Identify key voice and packet data KPIs
2
1 | WCDMA Performance Optimization
WCDMA Optimization
Principles and KPIs
3
1 | WCDMA Performance Optimization
Optimization Goals
Minimize
Latency Maximize Data
Maximize Voice Throughput
Quality
Minimize Maximize
Dropped Calls System Capacity
Maximize RF
Minimize Access
Capacity
Failures
The process of RF engineering and optimization has • System Capacity: The capacity of the system may be
multiple benefits, to both operators and subscribers: constrained at several points in the network, including
available downlink power, available codes, overhead
• Access Performance: Reducing access failures allows
channel capacity, and the capabilities of the network
more calls to be set up successfully, which translates
and transmission equipment. The traffic presented to
to both increased customer satisfaction and increased
the system must be taken into account when
minutes of use (MOU) on the system.
determining and optimizing the capabilities of the
• Dropped Calls: Similarly, dropped calls negatively radio network.
impact MOU and customer satisfaction. Call drops are
• RF Capacity: Also termed spectral efficiency, this is
typically affected by coverage and the amount of
the chief goal of the optimization effort. By increasing
dedicated resources used by the call.
RF capacity without adversely impacting the other
• Voice Quality: The quality of a mobile voice areas listed above, more users can be given better
conversation is inherently limited by the voice codec service.
being used. However, managing the amount of
The bottom line is to increase the potential of the radio
dedicated resources in order minimize errors over the
network to generate revenue for the service provider,
air can maximize the potential voice quality.
while balancing the performance of each of these areas.
• Packet Data Throughput and Latency: Packet data
performance differs from voice in that the typical
events that adversely affect voice users (i.e. access
failures and drops) have relatively little effect on most
data applications. However, these events do reduce
average throughput and increase latency.
4
1 | WCDMA Performance Optimization
Define Criteria
• Priorities – capacity Measurements
or quality? • Scanner Logs
• System Limits – RF or • Voice: Short/Long Calls
capital • Data: Small/Large Files
• Define KPIs • System Logs
• Set KPI Targets
Before embarking on an optimization effort, the network should be changed at any time in order to resolve a
priorities must be defined. problem, in order to minimize the “ripple effect”, but this
is not always practical.
• Is this a new network and quality is paramount?’ OR
Is this a mature network and quality must be
compromised to support capacity?
• How much capital is available to overcome network
issues?
An optimization program starts off with identifying the key
KPIs to be monitored. Target values for these KPIs are set
to meet the required performance. Measurements can
then be made to track these KPIs. When a KPI falls out of
the “good” range, action must be taken to identify and
address the underlying cause through an optimization
process. Optimization is the most challenging aspect of
the entire process. While some issues identified during
the measurement stage may be obvious, others may
require a detailed investigation and analysis to determine
the root cause. The scope and scale of the solution may
also vary widely, ranging from simply adjusting a
parameter value, to reconfiguring entire portions of the
radio network. Ideally, only one aspect of the system
5
1 | WCDMA Performance Optimization
Let’s discuss the factors that are the foundation to between the two networks is inevitable and the borders
optimizing performance in a WCDMA network. These must be optimized.
include:
In a system as complex as a mobile wireless network,
• Coverage system performance is a very dynamic thing: subscribers
move through the network, environmental conditions
• Neighbor Lists
change, and any adjustments usually have side effects.
Both of these items can have positive or negative RF optimization is an ongoing, iterative process that never
implications on all areas of WCDMA performance. That ends.
being said, we should always begin our process by
optimizing the coverage and neighbor list of the area of
interest. In this manner, we can achieve the most “bang
for our buck”. Once a network’s coverage and neighbor
list has be optimized, the parametric optimization phase
can begin. We first outline the approach to measuring and
improving each of these factors, then move into per-KPI
optimization. An iterative process of measuring, modifying,
analyzing each KPI is required.
Another aspect of WCDMA systems that should be
considered in the management and optimization of
WCDMA networks is inter-RAT mobility. Since WCDMA
networks coexist with GSM/GPRS networks and often may
not cover 100% of the area covered by GSM, mobility
6
1 | WCDMA Performance Optimization
Voice KPIs
Retainability
• Call drop rate Quality
• Handoff failure rate • Delay
• Jitter
• Call Setup Time
Resource Usage
Accessibility • Sectors per user
• Setup failure rate • Channel elements per user
• Paging success rate • Blocking rate
Key Performance Indicators (KPIs) can vary widely, receiving voice packets over the air (which correlates
depending on what information is available from the directly to what the user hears), while the call setup
equipment and on what is important to the operator. time assesses how long it takes the system to connect
Nonetheless, they tend to fall into a few broad categories. the user and begin a conversation.
For circuit voice services, the most commonly used KPIs
• Resource Usage: Although subscribers rarely have any
are:
visibility into the resources they use in their calls, it is
• Accessibility: The ability of a subscriber to get access of vital importance to the operator, since a lack of
to the network to make or receive a call. Setup failures resources means fewer users (and less revenue) and
track how often a user tried to set up a call, but was may require a significant investment in new
unable to do so, while paging success indicates how equipment. Tracking sectors and channel elements
often a subscriber in the area could be reached. per user allows the operator to assess how efficiently
they are serving their subscribers while the blocking
• Retainability: The ability of a subscriber involved in a
rate indicates how many users are adversely impacted
voice call to stay up and running. Call drops are the
by a lack of resources.
most visible (and annoying) aspect of retainability,
while handoff failures indicate problems in
maintaining connections as the user moves through
the network.
• Quality: Although every subscriber can easily evaluate
the quality of their phone call, it’s more challenging to
measure quality statistically. The block error rate
indicates how successful the system is in sending and
7
1 | WCDMA Performance Optimization
Scheduling
• DTX/DRX rates
• Traffic Levels
Quality
Resource Usage •Block Error Rate
• Allocated power • Latency
• Codes per user
In addition to Accessibility and Retainability
Award Solutions Proprietary
For packet data applications, the key metrics are a bit transmission rates associated with packet data
different, since the things that adversely affect voice calls require more resources (power and codes). A small
may not have the same impact on data users. number of data users can easily consume all of the
resources in a cell, completely blocking other users
• Throughput: For the majority of data applications,
(voice and data).
throughput is the primary concern; the faster a web
page is downloaded, or the quicker an e-mail is • Scheduling: Related to resource usage, the data
retrieved, the happier the subscriber is. Throughput is scheduling functions can have a dramatic impact on
often tracked at two levels: application throughput overall performance. In particular, the high-speed data
(which is what the user experiences) and physical solutions in UMTS, HSDPA and HSUPA, share the
layer throughput (which is what actually happens over resources among all of the active users. The scheduler
the radio link). in the Node B must ensure that each user gets his fair
share; excessive discontinuous transmission or
• Delay: The other key factor for data is delay: how long
reception (DTX or DRX) of data means that the user
its takes a packet to be delivered. Most common
will experience lower average throughput and higher
applications (like web browsing, file transfer and e-
average delay. High levels of data traffic may require
mail) are delay-insensitive: an extra few milliseconds
additional radio channels and/or cells in order to
make no discernable difference to the result. Real-
handle the load.
time services like Voice over IP (VoIP), however are
impacted, if there is too much latency and jitter
(variations in delay).
• Resource Usage: Resources are arguably a greater
concern for data than voice, since the higher
8
1 | WCDMA Performance Optimization
Summary
• Well-defined Key Performance Indicators (KPIs) drive the optimization
process.
– Accessibility, Retainability, Quality and Resource Usage are needed
for voice.
– Accessibility, Retainability, Throughput, Delay and Resource Usage
are needed for packet data.
• Optimization is an iterative process.
– KPI targets identify problem areas that require attention.
– Analysis and troubleshooting determine the root cause (or causes)
of the problem.
– Optimization finds the best approach to address the issue
(parameter changes, hardware upgrades, RF coverage adjustments,
etc.).
– KPIs are monitored after the change to ensure that the targets are
now achieved.
9
1 | WCDMA Performance Optimization
Review Questions
1. What is the purpose of network optimization?
2. What are the key metrics to be monitored to measure
performance of packet data applications?
3. Which of the key KPIs are affected when users’ experience
delay in call setup?
4. Which of the key KPIs are affected when users’ experience
more call drops in the network?
5. State True or False. Coverage and Neighbor lists are the
foundation to optimizing performance in a WCDMA
network.
10
2 | RF Coverage and Neighbor List Tuning
Chapter 2:
RF Coverage and
Neighbor List Tuning
11
2 | RF Coverage and Neighbor List Tuning
Objectives
After completing this module, you will be able to:
• Define the key RF-related measurements
• Explain the coverage optimization process
• Describe pilot pollution and the techniques to fix
it
• Explain the importance of optimizing neighbor
lists and its impact on the composite neighbor
list
• Detail the interpretation of key coverage plots
12
2 | RF Coverage and Neighbor List Tuning
RF Measurements
13
2 | RF Coverage and Neighbor List Tuning
RF Measurements
14
2 | RF Coverage and Neighbor List Tuning
Eb/No Ec/No
•Eb: strength of desired •Ec: RSCP of the common
signal - variable pilot channel - fixed
•No: strength of interfering •No: RSSI of the entire
signals channel
•Has benefits of Power •No benefits of power
Control, Soft Handover, control, Soft Handover,
CDMA Processing Gain CDMA Processing Gain
•Provides a measure of •Provides a measure of
signal quality relative path loss
Ec/No ≠ Eb/No
Award Solutions Proprietary
Although Ec/No is often interpreted as a measure of • No is the received strength of any interference that
channel quality, in reality, it means something quite affects the desired signal (Eb), and does not include
different. the desired signal itself
• Ec is the power (RSCP) of the common pilot channel, The ratio of these two values indicates how strong the
which is typically 10% or less of the total channel desired signal is, relative to its interferers.
power, has no power control applied to it and is not
impacted by any soft handover activity
• No is the power (RSSI) of the entire 5 MHz channel,
including the pilot (Ec), all other channels in the cell,
and all external interferers
The ratio of these two values reflects the relative strength
of the pilot, which correlates to path loss, not channel
quality. Comparing Ec/No measurements is useful for
determining handover events, but a low Ec/No does not
necessarily mean that the user is experiencing poor
performance.
Eb/No (or SIR), on the other hand, correlates well with the
overall quality of the signal:
• Eb is the received signal strength of the desired
signal i.e. a traffic channel with power control plus
any additional improvement through Soft Handover
15
2 | RF Coverage and Neighbor List Tuning
Ec No
Ec/No
A Ec/No B
16
2 | RF Coverage and Neighbor List Tuning
Coverage Optimization
17
2 | RF Coverage and Neighbor List Tuning
A WCDMA cell’s coverage is defined to be the area over Downlink CPICH-Ec/No is not valid for evaluating cell
which the uplink can be closed, where “closed” means coverage. CPICH-Ec/No says nothing about the likelihood
that uplink synchronization can be maintained. The uplink of closing the uplink. Also, even in the context of downlink
“in-sync” or “out-of-sync” status is determined by the closure, the CPICH is not power controlled, whereas the
uplink pilot bits strength Ec/No as measured by the BTS. downlink DPDCH is. Finally, Ec/No does not change
Since the pilot bits are transmitted on the DPCCH – not linearly with path loss, which any coverage metric must.
the DPDCH – the sync status, and therefore cell coverage,
are not strictly related to bearer data rate.
Downlink limits to cell coverage are artificial, i.e., the
result of parameters, not physics. All UTRAN products are
equipped with parameterized limits on the amount of
power that can be allocated to a given downlink radio
bearer. These parameters exist for capacity reasons, not
coverage reasons. If these parameters did not exist, or if
they were set at their maximum values, then the downlink
could be closed at any point in the system where the
uplink is closable. Thus, from the physics perspective, cell
coverage is strictly uplink-limited.
18
2 | RF Coverage and Neighbor List Tuning
19
2 | RF Coverage and Neighbor List Tuning
When tuning coverage, it is important to create the right similar to the above could ensue.
coverage at the right location. The right coverage includes:
“The more signal, the better” is not a good coverage
• Elimination of coverage gaps: As coverage is tuned, it strategy in any cellular technology, but it can be
is important not to open coverage holes. devastating in WCDMA. Resist adopting such a strategy.
• Minimization of pilot pollution: Although definitions
vary, pilot pollution can be described generally as the
state in which many pilots are received at a location
at very roughly the same strength. For the purposes
of this course, we usually use the following definition:
A point at which there are three or more additional
servers received within 5 dB of the strongest server is
deemed to be polluted. Pilot pollution can result in
excessive handoff, reduced cell capacity, increased
access failures, and dropped calls.
• Creation of clearly-dominant servers: Even when an
area does not satisfy the clinical definition of pilot
pollution, it could suffer from poor server dominance,
in which the best server changes rapidly from location
to location. This situation will be minimized by
ensuring that the best server clearly dominates all
other servers. Otherwise, performance problems
20
2 | RF Coverage and Neighbor List Tuning
UE
UE Node B
21
2 | RF Coverage and Neighbor List Tuning
Pilot Pollution
22
2 | RF Coverage and Neighbor List Tuning
Pilot Pollution
Why do we care about How do we search for
pilot pollution? pilot pollution?
Access failures, poor Definition: Three or more
quality, drops additional pilots within 5 dB of
strongest
Resource usage: power, Example:
codes, channel elements
• Best server CPICH Ec/No: -6 dB
• Pollution threshold: -6 – 5 = -11
dB
• Second, third, and fourth pilots
at -6.5, -7.5, and -9 dB
• The area is considered polluted
23
2 | RF Coverage and Neighbor List Tuning
UE
24
2 | RF Coverage and Neighbor List Tuning
A number of approaches to the coverage optimization A best server Ec/No plot might be useful in identifying
problem exist. One methodical way is as follows: polluted and/or poor dominance areas that haven’t
otherwise been revealed through the above tasks.
1. The first task is to confirm that minimum uplink
However, note that the task when inspecting an Ec/No
coverage levels have been met. We do this by
plot is not to ensure that Ec/No is above an arbitrary
ensuring that the best server RSCP is above a certain
threshold, but to search out areas of dramatic, relative
threshold, as determined by the uplink link budget.
drops in Ec/No that are indicative of excessive
2. The next task is to inspect the coverage of each cell interference. However, one must be careful when utilizing
by inspecting the best server Selected Cell (SC) plot. Ec/No plots. A cell serving users with either dedicated or
This can expose instances of over-propagation and HS channels will effect Ec/No values.
poor server dominance.
A natural side-effect of this process is a reduction in soft
3. Next, a pollution plot can come in handy. This plot handoff, which, in turn, will result in increased capacity of
indicates the presence, degree of severity, and the radio network.
perpetrators of pilot pollution so that geographic
areas can be targeted for optimization.
4. Next, a relative rank plot can be useful. A relative
rank plot is a per-cell plot that indicates the areas
where the SC is best server, second-best server, third-
best, and so on. This plot provides tremendous
insight into how well a cell dominates its indented
service area, and how badly it propagates out of its
intended footprint.
25
2 | RF Coverage and Neighbor List Tuning
• Tilts
Use RF • Azimuth changes
engineering • Antenna height
to correct • Antenna type
coverage
• New site
issues
• Disable cell
Correcting coverage issues is achieved by utilizing good • Disable Cell: Although this sounds counter-productive,
old-fashioned RF engineering. The tools available include in extreme cases, disabling a site may improve
the following: system performance. If an enabled site is duplicating
the coverage of another site, and the net capacity
• Antenna Tilts: This includes both mechanical and
gain is less than or equal to zero, disabling a site
electrical. Tilts can be used to correct over
should be considered. However, this assumes all
propagation, or coverage holes.
other avenues of RF optimization have been
• Azimuth Changes: A standard 0-120-240 azimuth considered.
configuration can result in multiple RF coverage
issues in a WCDMA network. Antenna azimuth
changes may alleviate RF coverage issues.
• Antenna Height: Although construction and lease
issues may make this option prohibitive, excessive
antenna height is a typical source of pollution.
• Antenna Type: The beamwidth and gain of antennas
will obviously impact RF coverage. This option may be
limited in some cases due to cost, or antenna sharing
requirements.
• New Site: In areas of limited coverage, or lack of
dominance, the best option may be a new site.
26
2 | RF Coverage and Neighbor List Tuning
Neighbor Lists
27
2 | RF Coverage and Neighbor List Tuning
The original neighbor list is created based on three better the results.
different inputs:
A combination of all three methods described above is
Geography: Initially, the geographical location of other typically sufficient input to create an initial neighbor list for
cells, relative to the a cell in question, is used to generate a cell. Once the cell is operational, the neighbor list can
its neighbor list. Normally, those cells within close be further refined with drive test data and actual network
proximity of a given cell are selected first. This obviously data collected via the OSS system.
includes co-located Node B cells. Since this method is
ineffective at identifying strong and distant cells, it cannot
be relied upon alone and must therefore be used in The alternative is to use Automatic Neighbor Relation
conjunction with other sources of data when planning a (ANR) tools to automatically detect the suitable neighbors.
new cell’s neighbor list.
2G Legacy Data: If a Neighbor List (NL) exists for an
overlaid 2G cellular system, the 2G NL could be used as
the initial 3G NL. This saves time, effort and money but is
only effective if a one-to-one overlay deployment is being
carried out.
RF Prediction Tools: RF simulation tools permit the
assessment of the RF environment by means of a Monte
Carlo simulation that provides information on the best
neighbors for each cell in the network. The more accurate
the assumptions and the data used in the simulation, the
28
2 | RF Coverage and Neighbor List Tuning
The Active Set is the group of cells that have allocated measured value meets the Soft Handover (SHO) addition
dedicated resources to a UE, while it is in the CELL_DCH or replacement criteria, then a Measurement Report
state. All cells in the Active Set transmit a dedicated radio indicating this is sent to the RNC.
link in the downlink to the UE, as well as attempt to
Finally, the Detected Set is comprised of any pilots the UE
receive uplink data broadcast by the UE. The UE may or
measures, that are not in the current Active or Monitored
may not allocate RAKE receiver resources (i.e., a RAKE
Sets, yet meet the criteria to be added to the Active Set.
finger) to all Active Set radio links at all times. The
Detected Set reporting by the UE is optional and is
instantaneous assignment of a RAKE finger to a specific
enabled by the RNC. This type of data can be utilized to
cell’s radio link, or its multipath, is determined by the
help identify missing neighbors.
RAKE’s Searcher Finger algorithm and is based on current
RF conditions. However, it is important to note that the
Searcher Finger can quickly allocate RAKE receiver
resources to an unused Active Set radio link if RF
conditions change.
A cell’s Monitored Set is typically referred to as its
neighbor list. This is the list of cells that are potential
candidates to be added to the UE’s Active Set, due to
geographical location or RF signal strength. The Monitored
Set is sent to the UE via Measurement Control messages
and is updated every time there is a change in the Active
Set. The UE’s Searcher Finger periodically measures the
pilot strength of the cells in the Monitored Set. If this
29
2 | RF Coverage and Neighbor List Tuning
As the UE takes an active call through a WCDMA network, so that the strongest cell or reference pilot’s neighbors
the RNC or UTRAN is responsible for ensuring that the UE are added to the CNL first, then the second-strongest
is equipped with a suitable neighbor list that can ensure Active Set members neighbors, and so-on. This is based
good call or data performance through the entire duration on the knowledge of the UTRAN and the strength of each
of the call or data session. When there is only a single cell Active Set member, provided by the UE in the most recent
in the Active Set, the Monitored Set is obviously the Measurement Report message received. Of course, the
neighbor set of that specific cell. However, when multiple UTRAN considers duplicate neighbors amongst the various
cells are in the active set, there must be a mechanism for neighbor lists before constructing the CNL.
the RNC to create a Monitored Set that is derived from the
individual neighbor lists of the Active Set cells. This
blended neighbor list is the Composite Neighbor List
(CNL). It is also important to understand how this
combining is performed and that it can vary from
infrastructure vendor to vendor since it is the result of a
proprietary vendor algorithm. This could result in two
vendors potentially performing slightly differently in the
same RF environment and given the same initial neighbor
sets to combine.
3GPP does not specify how to construct the CNL, so
algorithms can vary between vendors. Most algorithms
take into consideration the strength of the donor cell (the
cell providing the neighbor list or the Active Set member),
30
2 | RF Coverage and Neighbor List Tuning
RF Coverage Control
It is important to understand that the neighbor list in system data that can be used to reevaluate the initial
WCDMA systems should be a direct reflection of the RF neighbor lists and begin optimizing these as we will
environment. Unlike GSM networks, neighbors cannot be discuss in the next slide.
added based on geography only. Given the nature of
As stated earlier poor RF control situations or boomer cells
mobile wireless environments each sector’s neighbor list
are the primary source of duplicate SCs occurring in CNLs,
must be tuned to accurately reflect the network RF
resulting in negative performance impacts to the system if
conditions. Which could result in neighbor candidates
not caught and excluded by the RF engineer and/or
being assigned from several tiers away. In these
infrastructure vendor CNL algorithm. These unwanted
instances it is the RF engineers responsibility to try and
and sometimes unavoidable RF control issues make the
modify these RF conditions so that unwanted neighbors
task of scrambling code planning very challenging and
are significantly reduced in number and preferably
time consuming. For this reason many of the latest RF
eliminated wherever possible. However, until this can be
simulation tools now have the ability to also help define
done the engineer must take care to include these
the network SC plan. The advantage of this is that these
boomers in the relevant sectors neighbor lists and
tools can quickly and exhaustively analyze all possible CNL
perform the analysis required to ensure that any resulting
scenarios based on the proposed SC plan and either
composite neighbor lists will not result in any situations
eliminate or identify any situations at risk of causing a
where multiple neighbor cells are using the same SC.
duplicate SC condition in the CNL. A task that if done
So what this means is the initial neighbor lists for a manually would have most RF engineers wishing that they
system design may be longer than desired. This should could go back to frequency planning. What is important to
help to avoid any missing neighbor issues near term and note however is that most of these tools cannot account
as the system matures the engineers will have time to for unique situations such as in-buildings, tunnels or at
properly address any poor RF control issues and collect high elevations in or near dense urban environments.
31
2 | RF Coverage and Neighbor List Tuning
32
2 | RF Coverage and Neighbor List Tuning
Although NLs can contain as many as 32 members, their result of the NL size reductions. Thus, creating issues the
length should strike a balance that ensures complete RNC or UTRAN will be incapable of correcting since the
coverage of all valid neighbors and minimal length to missing neighbor SCs are no longer included in the
improve the NL search speed and efficiency. The longer Monitored or Active Sets, as they may have been in a
the neighbor list, the less frequently each neighbor can be previous truncated CNL case. There is a delicate balance
searched. This can be problematic in dense urban, small between ensuring the NL is long enough to include all
cell radius, or high-mobility environments where pilots can valid neighbors, while simultaneously ensuring the NL is
change frequently due to rapidly changing RF conditions. short enough and prioritized correctly to minimize search
This can result in situations where active and neighboring times. Either of these situations could result in a sudden
pilot signal strengths change rapidly between successive drop in signal quality or degradation in call quality that
searches of the pilots, and, if the priority and length of the could have been avoided with properly optimized NLs. This
neighbor lists is incorrect, there is a chance that call also assumes that the RF environment has been well-
quality may degrade quicker than the neighbor and Active optimized to minimize or avoid creating any pilot pollution
Set members can be detected and promoted or demoted or non-dominant server conditions throughout the
accordingly. In addition, the longer the neighbor list, the network, as this is a prerequisite to almost any
higher the risk of causing a truncated CNL situation. optimization of a WCDMA network.
Short NLs are preferred since they help optimize search
times, reduce the chance of excluding neighbors in the
CNL and minimize or, preferably, eliminate any
dependence on the Detected Set. The caveat, however, is
that making NLs too short will only serve to create poor RF
conditions where missing neighbor issues could exist as a
33
2 | RF Coverage and Neighbor List Tuning
Inputs Tasks
NL tuning can be based on network wide scanner drive antenna heights, azimuths, downtilts or types, or
test data. However, this is not typically a very good increased SHO leg support to help reduce or compensate
representation of the actual subscriber usage and for pilot pollution issues causing a lack of dominant
therefore cannot identify all of the significant RF servers in areas of the network identified by this data.
interference issues. So in addition to drive test data, and One way to use this Measurement Data to automatically
potentially in place of it long term, neighbor list tuning develop the neighbor list is through “Automatic neighbor
must also be done using data collected at the RNC/OSS or relation (ANR)” tools.
via third party tools that capture a broader range of
network requirements drive tests cannot account for.
The UTRAN should be capable of supporting the logging or
recording of reports containing the content of the
Measurement Report messages from the UEs. This data
is invaluable for neighbor list tuning since it is based on
actual UE measurement reports and therefore is
considerably more representative of real subscriber usage
and traffic patterns throughout the network. By
accumulating an historical database of this information
the operator will now have the vital data needed to
optimize their NLs as well as identify potential areas of the
network where interference may exist due to RF control
issues. This data can then be used by the RF engineers to
propose changes such as the addition of new cells, modify
34
2 | RF Coverage and Neighbor List Tuning
In addition, there are other rules that must be observed Additional rules for neighbor planning related to actual SC
while creating and tuning a neighbor list: used must also be considered, including the following:
• The co-sited cells or adjacent sectors of a single site • No neighbor can use the same SC as the host cell SC.
must always be added as neighbors to each other,
• No two neighbors in the same individual neighbor set
regardless of whether they appear in scanner logs or
can use the same SC.
not.
• No two neighbors of a CNL can use the same SC.
• The reciprocity rule should be observed but may have
exceptions, so be cautious with this rule. If X is added This last rule pertaining to the assignment of SCs is why
as a neighbor of Y due to scanner logs, then Y should careful attention must be paid to any neighbor candidate
be added as a neighbor of X. that may be included in a serving cell neighbor list that is
more that two tiers away from the serving cell. This is also
― A possible exception here could be if X is a
why the reciprocity rule may not always be appropriate.
distant cell that propagates over Y and is a valid
When combining neighbor lists from distant cells, the risk
neighbor to Y but Y is not a valid neighbor of X
of having a duplicate SC condition becomes higher and SC
since it does not have the same RF propagation
planning becomes much more difficult.
concerns and cannot be seen as having a UE
camp or originating a call in Y’s coverage area. If
this can be confirmed, then Y should not be
included in X’s neighbor list. Now, to be safe
when initially deploying a new network or site, it
is probably wise to use the reciprocity rule until
better data is available to make such decisions.
35
2 | RF Coverage and Neighbor List Tuning
Dropped calls
Increased UL Interference
•Other cell dropped calls
•Reduced capacity
Incorrect neighbor lists are one of the major issues faced 3. Dropped Call Rate: When the interference of a
in performance optimization of WCDMA networks. missing neighbor gets high enough that power control
Neighbor list issues are often found to be the source of or can no longer compensate for it, a call drop will
a contributing factor to negative impacts on: typically result.
1. Access Success Rate: The UE unable to select to the 4. UL Interference: Since a missing neighbor is not part
correct best server due to aliasing or a missing of the UE’s Active Set, it will not be able to power
neighbor. As a result, the UE is not attempting to control that UE. This will result in the RTWP of the
access the best server and may not be able to missing neighbor cell to increase. This, in turn,
overcome the path loss to the this cell, or overcome requires all UEs in that cell to increase their power to
the interference from cell the UE should have been compensate, which contributes to the overall UL
accessing. interference level to increase. The end result being an
overall reduction in the system capacity and
2. Voice Quality or Data Throughput: A strong missing
throughput performance.
neighbor prevented from being added to the Active
Set becomes a source of destructive co-channel Thus, a simple missing neighbor can result in a cascading
interference for a UE already in a call or data session. domino effect that potentially may be seen as an overall
If the interference is strong enough, it could result in system or cluster performance issue. The problem with
causing a dropped call, but, at a minimum, it will likely this is that now the missing neighbor, or root cause of the
cause the voice quality or data throughput to problem, is being masked from easy detection. This could
degrade, or dedicated power resources to increase. prevent a timely, cost-effective solution from being
identified.
36
2 | RF Coverage and Neighbor List Tuning
Dropped calls
Increased UL Interference
•Other cell dropped calls
•Reduced capacity
37
2 | RF Coverage and Neighbor List Tuning
Summary
• WCDMA cell coverage is uplink-limited, however, most
problems are a result of the downlink coverage.
• Coverage optimization is mostly about controlling downlink
(over-) propagation:
– Minimizing pollution, and
– Maximizing server dominance
• Good RF control and thorough NL validation are the keys to
ensuring a proper NL design.
• NLs must be complete first, then optimized for efficiency
and performance.
• Historical system usage data is the key to achieving an
optimal NL design and ensuring long-term maintenance.
38
2 | RF Coverage and Neighbor List Tuning
Review Questions
1. Which of the RF measurements is used by the UE to
trigger handover events?
2. Which of the RF measurements is used to measure signal
quality in WCDMA network?
3. What is meant by “Right Coverage” ?
4. What is pilot pollution? What is the impact of it on
network performance?
5. What are the inputs considered while creating initial
neighbor list for a cell?
6. How incorrect neighbor list tuning impacts on network
performance?
39
40
3 | Access Performance
Chapter 3:
Access Performance
41
3 | Access Performance
Objectives
After completing this module, you will be able to:
• Explain the call setup process
• Describe the random access procedure
• Analyze the parameters that affect access
performance
• Define access success and failure
• Explain the modes of failure for access
• Describe the methods of resolving access issues
42
3 | Access Performance
• Characterize failures
Tasks • Identify performance impacting
parameters
43
3 | Access Performance
Retainability Phase
There are two distinct stages of call setup one should First, the UE is assigned an appropriate Radio Access
consider when studying WCDMA access performance – Bearer to carry the service being setup. In our examples,
also referred to as the Accessibility Phase: this service is speech using the AMR vocoder. Second is
the NAS signaling [Alerting, Connect, Connect Ack] to
• RRC Connection Setup
finish the call setup.
• Radio Access Bearer (RAB) Setup
Retainability Phase starts after the successful setup of the
In the first stage, the UE is taken from idle mode, through RAB i.e. completion of Accessibility Phase. Any call failures
the random access process to assignment of an RRC in the Retainability Phase are referred to as dropped calls
connection to establish 4 Signaling Radio Bearers i.e. calls complete through an abnormal release.
(SRBs)for AS ands NAS signaling. For the examples and
case studies in this module, it is assumed that the UE is
placed in Cell_DCH state at the close of this stage of the
call setup. Referring to the earlier call flow diagram, the
first stage begins with RRC Connection Request and ends
with the UE sending an RRC Connection Setup Complete.
At this point, the UE is on the dedicated physical channel
and has been assigned 4 SRBs to carry out the remainder
of the call setup.
In the second stage of the call setup, beyond
authentication and security, there are two key steps that
take place.
44
3 | Access Performance
Random Access
Process and
Parameters
45
3 | Access Performance
RACH Overview
• Minimize the chance of collision with
other UEs trying to access the network
• Provide enough coverage to insure that
Goals the RACH can be sent from all coverage
areas of the cell
• Minimize transmitted power to reduce
interference to all other UEs
We now take a more detailed look at the first stage of call The goal, then, of this process is to minimize the chance
setup: the random access process. The UE starts the call of collision and minimize the initial UE transmit power so
setup in the idle mode. As with all WCDMA operations, the that interference to other UEs is reduced. Both of these
UE must first establish an RRC Connection before any goals are in place to increase the reliability of the process
service can be initiated. and to shorten the call setup time.
The UE utilizes the uplink Random Access Channel (RACH)
to begin this process. There are several facts that must be
understood when looking at the goals of the RACH
process:
• The uplink PRACH/RACH is a shared medium that
may encounter contention during multiple users’
access attempts;
• All UEs transmit on the same uplink WCDMA channel
and therefore interfere with each other; and
• Each UE must estimate the amount of power to use
on the access attempt since no feedback from the
Node B exists as it does on a dedicated channel.
46
3 | Access Performance
RACH Procedure
Power powerOffsetPpm
Time
Preamble Preamble Preamble N = preambleRetransMax
1 2 3 (Max number of preamble Txs allowed)
One Preamble Cycle, repeat mMax times
When an Idle UE requires an RRC connection, the UE’s If an Acknowledgement (ACK) to this preamble is received
RRC Layer will initiate the RRC Connection procedure, on the Acquisition Indication Channel (AICH), the UE will
indicating a message needs to be sent on the RACH. In transmit the RRC Connection Request message. The
order to minimize uplink interference, and reduce power used to transmit the RRC message is determined
potential collisions with other UEs accessing the RACH, by adding the value powerOffsetPpm to power used by the
this procedure utilizes open-loop power control. The UE, last transmitted preamble.
prior to sending the RRC Connection Request, determines
However, if the transmitted preamble does not receive an
the initial transmit power through parameters broadcast
acknowledgement by the AICH channel, a second
via System Information messages, and measurements of
preamble will be sent with its power increased by the
that target cell’s pilot channel. The initial UE transmit
parameter powerRampStep. This process will continue
power is derived from the open-loop power control
until either an AICH acknowledgement is received, the
equation: Initial UE Transmit Power = primaryCPICHpower
maximum UE transmit power it reached, or the maximum
– CPICH_RSCP + UL_interference + constantValueCprach
number of transmitted preamble limit –
This equation estimates the minimum power required to preambleRetransMax - has be reached. At this point, one
close the uplink for the RRC Connection Request preamble cycle has be completed.
message. This is based on downlink path loss
After a random backoff time, the MAC Layer repeats this
(primaryCPICHpower – CPICH_RSCP), the uplink noise that
entire preamble cycle for a total of mMax cycles, thus
must be overcome (UL_interference), and the processing
completing one RRC RACH attempt.
gain due to spreading for the RRC message
(constantValueCprach). Once transmit power is If necessary, the RRC Layer will try a total of N300 times
determined, the UE’s Physical Layer will transmit a to have a successful RACH attempt, with T300 delay
preamble at this calculated power. between tries.
47
3 | Access Performance
RACH Parameters
System
SIB 1 – NAS and Information
UE Timers SIB 7 - RTWP
UL Interference
N300 SIB 5 – Common
T300 Physical Channels
• Available Signatures
• Available Sub-channels
• primaryCPICH-TX-Power
• constantValueCprach
• powerRampStep
• powerOffsetPpm
• preambleRetransMax
• mMax
Award Solutions Proprietary
48
3 | Access Performance
UL Interference
N300 SIB 5 – Common
T300 Physical Channels
• Available Signatures
• Available Sub-channels
• primaryCPICH-TX-Power
• constantValueCprach
• powerRampStep
• powerOffsetPpm
• preambleRetransMax
• mMax
Award Solutions Proprietary
49
3 | Access Performance
50
3 | Access Performance
The dynamic range of a UE is defined by its minimum and To mitigate this problem, three possible parameter
maximum transmit power. The minimum is specified by changes are discussed:
3GPP to be -50 dBm, while the maximum is operator-
1. Increase the value of contantValueCprach (e.g., from -
specified and typically set to 24 dBm. Therefore, the
27 to -15 dB). While this option will solve the problem
dynamic range for a UE’s transmit power is typically 74
discussed, it is not recommended. By grossly
dB. On the other hand, the dynamic range of the random
offsetting the initial preamble transmit power, the
access procedure is defined by the product of the
result will be excessive system noise rise in most RF
preamble step size (powerRampStep) and the max
conditions.
number of preambles (preambleRetranMax). For example,
if powerRampStep = 2 dB, and preambleRetransMax = 2. Increasing the step size of the Preamble
15, the preamble dynamic range will be equal to 28 dB. (powerRampStep). While this option will increase the
This is far less than that of the UE’s dynamic range. preamble dynamic range, it will also result in
Therefore, it is imperative that the initial preamble increased preamble overshoot. Preamble overshoot is
transmit power, defined by the open-loop power control defined as the amount of power the preamble
equation, is reasonably accurate. exceeds the required transmit power. Because the
average preamble overshoot is equal to half the step
However, in very good RF conditions, it has been observed
size, increasing the step size would contribute to
that the initial preamble transmit power can be
increased interference in the cell.
underestimated. As illustrated in this slide, this can result
in all of the preambles being transmitted without obtaining
the required transmit power required to reach the Node B.
51
3 | Access Performance
52
3 | Access Performance
Lack of Evidence
Why
won’t my
Great news,
Our network
call go
indicates no through?
failures. Random Access
Failures
53
3 | Access Performance
Access Failure
Scenarios
54
3 | Access Performance
55
3 | Access Performance
In order to troubleshoot and optimize accessibility issues, synchronization, a drop will occur. If this drop occurs
it is critical that the engineer understands where issues before the defined KPI threshold for a successful access
may arise as the call is set up. (e.g., Radio Bearer Setup Complete), it may be pegged as
an access failure. Regardless, the RF engineer must be
During random access, it is imperative that the UE has the
able to recognize this type of access failure as a drop, and
ability to close the uplink. This may not be possible due to
treat it as such.
RF conditions (e.g., lack of coverage), or excessive UL
noise impairing the target cell. In addition, access issues
are likely to occur if the UE is not accessing the correct
cell. Incorrect neighbor lists, pilot pollution, and
overshooting cells can lead to the UE being camped on
the wrong cell.
Once an RRC connection is established, and the UE
continues the call setup process, core network issues and
resource contention may prevent the call from being set
up. These sort of issues are typically easy to identify with
counters and network traces.
Finally, good RF coverage and a well-tuned neighbor list
will improve the likelihood of a successful call setup. Once
the UE has established a dedicated channel, it can use
soft handover and power control to mitigate interference.
However, if both the UE and network cannot maintain
56
3 | Access Performance
Summary
• Access process is divided into two stages:
– RRC connection setup and
– Call completion on DCH.
• A variety of factors can affect access performance
including:
– Random access,
– Admission control,
– Core network issues, and
– Uplink and downlink coverage (interference, incorrect
neighbor lists).
• Recognize the difference between an access failure and a
drop.
57
3 | Access Performance
Review Questions
1. If network statistics data show RRC connection setup
success rate as 100 percent and RAB setup success rate
as 92 percent, which of the KPIs are affected by this?
2. What RACH parameters you prefer to change to increase
preamble dynamic range?
3. What could be the impact if you set N300 counter
(maximum number of retransmissions of RRC Connection
Request message) to maximum value?
4. What could be the reasons for DCH drops prior to call
establishment?
58
3 | Access Performance
Additional Charts
59
3 | Access Performance
Identify Cell
Admission Parameter RF
Issues Issues Issues NL Issue
Drive Test
60
4 | Dropped Calls and Call Quality
Chapter 4:
Dropped Calls
and Call Quality
61
4 | Dropped Calls and Call Quality
Objectives
After completing this module, you will be able
to:
• Understand factors that can cause dropped calls
• Illustrate the modes of failure and outline the
methods of resolving dropped calls
• Summarize the value of the Layer 3 UE
Measurement Reports from RRC messages
• Describe DL BLER for voice calls
62
4 | Dropped Calls and Call Quality
• Characterize failures
Tasks • Identify performance-impacting
parameters and actions to resolve drops
63
4 | Dropped Calls and Call Quality
64
4 | Dropped Calls and Call Quality
65
4 | Dropped Calls and Call Quality
66
4 | Dropped Calls and Call Quality
67
4 | Dropped Calls and Call Quality
Downlink BLER
• Quality Target set for
– Signaling Radio Bearer
– Each of the 3 AMR 12.2 Transport Channels
• BLER Reports from UE
– Signaling Radio Bearer
– First (A-bits) of the 3 speech Transport Channels
0.1*QualityTarget = LOG10(DesiredBLER)
68
4 | Dropped Calls and Call Quality
Loss of Synchronization
• Dropped Calls
– At the physical layer, a dropped call is due to one
of two things
• Loss of UL Synchronization of all active set links
– Node B informs RNC which then starts the process of
removing radio links
• Loss of DL Synchronization
– UE releases the radio links and starts cell reselection
process and inform RNC of the failure
Ultimately, a dropped call is the result of the lost Understanding these mechanisms of a dropped call will
synchronization of physical frames in either the uplink or aid the engineer in quickly identifying if a drop was caused
downlink. by an uplink or downlink issue.
In the case of the former, the Node B monitors the
synchronization status of the UL radio link, and
communicates this status to the RNC. If all the radio links
in the active set lose synchronization, the RNC will initiate
a IU Release Request with the core network, which will
lead to a network initiated release of the call.
In the case of downlink synchronization, the UE monitors
the downlink physical frame’s synchronization status. In
the cases of the a UE in soft handover, all radio links are
combined in the rake receiver. Therefore, the
synchronization status of the combined downlink radio
frames are monitored. If synchronization is lost, the UE
will release the call and start cell reselection process.
Once the UE has reselected successfully to a cell, it will
notify the network of the release with a “Cell Update”
message indicating a “radio link failure.”
69
4 | Dropped Calls and Call Quality
UL RL Synchronization - I
• Radio Link Failure Indication
– The Node B checks the synchronization status of every 10 ms frame
– If N_OUTSYNC_IND consecutive “out of synch” indications are
received, timer T_RLFAILURE is started
– If N_INSYNC_IND consecutive “in synch” indications are received,
before T_RLFAILURE has expired, the timer is reset and the RL
remains in the in synch state
– If the timer expires, the Node B will put the RL in the “out-of-synch”
state and send a Radio Link Failure Indication message to the RNC
via the NBAP.
Award Solutions Proprietary
70
4 | Dropped Calls and Call Quality
UL RL Synchronization - II
• Radio Link Failure
– When the RNC receives a Radio Link Failure Indication
message, a timer will be started to allow for RL
resynchronization
• If Active Set > 1
– If timer expires, RNC may use Active Set Update to delete RL
• If Active Set = 1
– If timer expires, RNC sends an Iu Release Request message to the
CN and the call is released
• Radio Link Restore
– If the RNC receives a Radio Link Restore Indication
before the timer has expired, the RL will not be deleted.
• The Node B sends a Radio Link Restore Indication when
N_InSync_Ind consecutive “in synch” indications are received
after the RL has been put into the “out of synch” state
71
4 | Dropped Calls and Call Quality
DL RL Synchronization
• DL Radio Link Failure
– The UE RRC confirms DL RL Synch of every 10-ms frame of the
DPCCH
– If N313 consecutive “out of synch” indications are received, timer
T313 is started
– If N315 consecutive “in synch” indications are received, before
timer T313 has expired, the timer is reset and the RL remains in the
in synch state
– If T313 expires prior to receiving N315 consecutive “in synch”
indications, a Radio Link Failure will have occurred.
– When an RL failure occurs:
• Clear the dedicated physical channel configuration
• Send cell update with cause “radio link failure”
Award Solutions Proprietary
72
4 | Dropped Calls and Call Quality
73
4 | Dropped Calls and Call Quality
Dropped Call
Scenarios
74
4 | Dropped Calls and Call Quality
Interference
• Incorrect Coverage
Neighbor
Configuration • Max Power
Soft Handover
• Pilot Pollution (UL and DL) Restricted
• Fades
• Obstructions
75
4 | Dropped Calls and Call Quality
Interference Drops - I
Incorrect Neighbor Configurations:
Cell 3
Cell 1 Cell 2
76
4 | Dropped Calls and Call Quality
Interference Drops - II
Pilot Pollution:
77
4 | Dropped Calls and Call Quality
Coverage Drops
Max power allocated:
UE inside
building
Far away UE
78
4 | Dropped Calls and Call Quality
Cell 2
Cell 1
The UE is served by Cell 1, and
would benefit from SHO with Cell
2, but the network won’t allow it
79
4 | Dropped Calls and Call Quality
Cell 2 Coverage of
Cell 2
80
4 | Dropped Calls and Call Quality
Summary
• RF-related drops are due to interference, which
results in loss of either UL synchronization or
DL synchronization.
• Loss of UL or DL synchronization can result in
the inability to exchange L3 messages including
active set updates, radio bearer
reconfigurations, etc.
• Interference may be caused by RF coverage
holes (excessive path loss), pilot pollution
and/or missing neighbors.
81
4 | Dropped Calls and Call Quality
Review Questions
1. What are the possible reasons for call drop?
2. What is meant by ”Missing Neighbor?” How does this
impact call drop rate?
3. How do soft handover parameter settings impact the call
drop rate?
82
4 | Dropped Calls and Call Quality
Additional Charts
83
4 | Dropped Calls and Call Quality
Identify Cell
Admission Parameter RF
• Neighbor List
Issues Issues Issues NL Issue Tuning
• Coverage Constraints • IRAT
• Capacity • Neighbor List
• Pilot Pollution configuration
Congestion Tuning
• Broken/Weathered • Slow SHO
• Coverage • IRAT Drive Test
Constraints configuration antenna
• Specific UE performance
Award Solutions Proprietary
84
4 | Dropped Calls and Call Quality
85
86
5 | Inter-Frequency and Inter-RAT Performance
Chapter 5:
Inter-Frequency
and Inter-RAT
Performance
87
5 | Inter-Frequency and Inter-RAT Performance
Objectives
After completing this module, you will be able to:
• Describe the different scenarios, parameters and
setting options associated with idle-mode
reselection
• Analyze the configurations, triggers and parameters
used for inter-frequency and inter-RAT connected-
mode handovers
• Illustrate the options, benefits and concerns for
inter-frequency and inter-RAT UE device
measurements
• Define the signaling messages used to execute and
analyze the performance of IFHO and IRAT
handovers Award Solutions Proprietary
88
5 | Inter-Frequency and Inter-RAT Performance
Idle-Mode Cell
Reselection
89
5 | Inter-Frequency and Inter-RAT Performance
90
5 | Inter-Frequency and Inter-RAT Performance
Initiating Reselection
Measurements (CPICH Ec/No)
WCDMA f1 and f2
CPICH Ec/No (dB)
Measurements
-8 dB
Sintrasearch=10 dB
-10 dB
Sintersearch=8 dB
-14 dB
SsearchRAT=4 dB
Qqualmin = -18 dB
t (sec)
91
5 | Inter-Frequency and Inter-RAT Performance
Initiating Reselection
Measurements (CPICH RSCP)
CPICH RSCP (dBm)
SsearchHCS = 6 dBm
-109 dBm
-111 dBm
SHCS-RAT = 4 dBm
The CPICH RSCP triggers for beginning neighbor cell If the CPICH RSCP drops further to below -111 dBm
measurements are similar to the Ec/No case, with one (Qrxlevmin + SHCS-RAT), the UE also begins to measure any
exception: intra-frequency measurements can not be GSM neighbors. If the RSCP of the current serving cell
triggered by RSCP, only Ec/No. Since comparing the Ec/No climbs above -111 dBm prior to a reselection occurring,
of two intra -frequency cells is exactly the same as the UE will stop measuring GSM neighbor candidates.
comparing their RSCPs, there is no need for an extra Furthermore, if the RSCP improves to a value better than -
trigger. 109 dBm, inter-frequency measurements will cease as
well.
The parameters SsearchHCS and SHCS-RAT are offsets that
define the thresholds at which the UE will begin measuring Note that the HCS search parameters are only supported
inter and IRAT neighbors respectively, relative to the value on release 5 devices or later. Prior to this, only CPICH
Qrxlevmin. Ec/No could be used to trigger cell reselections.
As an example, the slide illustrates these thresholds base
on the following parametric values:
• Qrxlevmin: -115 dBm
• SsearchHCS: 6 dBm
• SHCS-RAT: 4 dBm
As the CPICH RSCP of the serving cell drops below -109
dBm (Qrxlevmin + SsearchHCS), the UE will begin measuring
inter-frequency neighbors. This is true even if the serving
cell’s CPICH Ec/No is greater than Qqualmin + Sintersearch.
92
5 | Inter-Frequency and Inter-RAT Performance
Inter-Freq Reselection
(F1 to F2 Example)
F2 CPICH Ec/No
Treselection Qoffset2sn = 7 dB
Treselection
Qhys2s = 2 dB
When to Reselect
Award Solutions Proprietary
Once neighbor measurements have begun, the target cell have a unique value for each neighbor relation. In this
must be ranked higher than the serving cell for a period manner, Qoffset can be used to bias reselection for a
of Treselection. The UE computes the ranking of the serving specific serving cell or neighbor relation. On the other
cell “s” as hand, Qhyst is simply a hysteresis value added to the
serving cell’s CPICH signal level. Qhyst is broadcast in SIB
Rs = Qmeas,s + Qhysts
3/4 and the same value is added to the CPICH signal level
The UE computes the ranking of the neighbor cell “n” as for all neighbor evaluations.
Rn = Qmeas,n − Qoffset ,s,n The network actually provides two sets of values. Qhyst1s
If multiple neighbor cells are ranked higher than the and Qoffset1sn are used for reselection based on CPICH
serving cell, the highest ranked cell will be selected. Since RSCP, while Qhyst2s and Qoffset2sn are used for reselection
the ranking of the serving and neighbor cells are based on based on CPICH Ec/No. If Qoffset is not broadcast in the
the parameters Qhyst and Qoffset, the value for these SIB, then a default value of 0 will be sued. In addition, if
parameters may be used to tune reselection for intra, Qhyst2s is not broadcast, the value of Qhyst1s is used.
inter, and IRAT target cells.
In the case of intra-frequency or inter-frequency, the Effectively the short message from this chart is:
CPICH signal level of the measured neighbor cell must
When the Ec/No of the target cell is better than the Ec/No
exceed the serving cell by (Qhyst + Qoffset) for a period of
of the camped-on cell by (Qhyst2s+Qoffset2sn) dB for at
Treselection. If these criteria is met, the UE performs a cell
least Treselection seconds, UE reselects to the target cell.
reselection operation and camps on the new cell.
Although it may appear that Qhyst and Qoffset serve the
same purpose, Qoffset is broadcast in SIB 11/12 and may
93
5 | Inter-Frequency and Inter-RAT Performance
Inter-RAT Reselection
(UMTS to GSM Example)
Target GSM Cell RSSI
Treselection Qoffset1sn = 7 dB
Treselection
Qhyst1s = 2 dB
94
5 | Inter-Frequency and Inter-RAT Performance
Inter-Frequency Reselection
Cluster (Hot Spot) Overlay
Idle Mode f1B f1B f2A
Reselection:
Sintersearch 8 dB 8 dB 8 dB
Qhyst1 2 dB 2 dB 2 dB
Treselection 1 second 1 second 1 second
Neighbor List F2A F1A F1B
Qoffsets1,s,n 3 dB 2 dB 1 dB
f2 A
UEB
UEA f1 B f1 A
This slide illustrates how Qoffset1sn can be used to manage UEA would eventually perform an intra-frequency
reselection between inter-frequency cells. In this example, reselection to f1A based on a Qoffset1sn of 2 dB.
f2A is a second frequency cell, co-located with f1A. The
• Scenario 2: UEB moves, in idle mode, on f2A toward
coverage of the second carrier cell f2A is greater than its
the coverage area of cell B, supporting only carrier f1.
co-located cell f1A. This would be typical in a non-
UEB will now transition from a multi-carrier cell to a
contiguous second-carrier deployment of capacity, or
single-carrier cell. In order to ensure that UEB quickly
coverage sites. Since the second carrier cell is exposed to
reselects to f1B before its serving cell f2A becomes
less intra-frequency interference (both UL and DL), its
unusable, the parameter Qoffset1sn is set to a lower
coverage relative to f1A may be greater.
value of 1 dB. This configuration ensures that UEB is
Because of this difference in coverage, a reselection camped on a cell that has the RF coverage to support
strategy between f1 and f2 must be employed. In this a call if required. This is especially true if reselection
example, the value of Qoffset1sn is tuned per cell to achieve is based on Ec/No, which may inaccurately measure
the desired effect. A number of reselection scenarios are quality at the edge of f2A.
possible in this environment:
• Scenario 1: UEA moves, in idle mode, on f1B toward
the coverage area of cell A, supporting carriers f1 and
f2. UEA will now transition from a single-carrier cell to
a multi-carrier cell. In order to prevent UEA from
reselecting to f2A prematurely, the parameter Qoffset1sn
is set to a higher value of 3 dB. Alternatively,
reselection from f1B to f2A can be totally prevented by
simply omitting f2A as a neighbor of f1B. In this case,
95
5 | Inter-Frequency and Inter-RAT Performance
Inter-RAT Reselection
GSM to UMTS
RSSI (dBm)
Qsearch_I
Qsearch_I = 7, Always
Search for 3G Cells
t (sec)
GSM WCDMA and GSM GSM
Measurements Measurements Measurements
When to Start Taking Measurements
Award Solutions Proprietary
While in GSM mode, the UE monitors the system Excerpt from 3GPP TS 45.008 regarding parameter
information to determine if UMTS cells exist and whether Qsearch_I:
they should be measured. The parameter Qsearch_I
Search for 3G cells if signal level is below (0 … 7) or above
specifies when to begin measuring UMTS neighbors. If the
(8 … 15) threshold
signal of the serving GSM cell drops below Qsearch_I (e.g.
-98 dBm for level 0), then the UE will begin searching for 0 = - 98 dBm, 1 = - 94 dBm, … ,
UMTS neighbor candidates. If the signal of the serving 6 = - 74 dBm, 7 = infinity (always)
GSM cell becomes equal to or better than Qsearch_I prior
to a reselection, the UE will stop measuring UMTS 8 = - 78 dBm, 9 = - 74 dBm, … ,
candidates. 14 = - 54 dBm, 15 = infinity (never).
Note that a value of Qsearch_I = 7 indicates that the UE Default value = infinity (never).
shall always search for a UMTS candidate, while the
default value of Qsearch_I (15) indicates that the UE shall
never search for a UMTS candidate. The Qsearch_I
parameter is transmitted in GSM System Information Type
2quater.
96
5 | Inter-Frequency and Inter-RAT Performance
Inter-RAT Reselection
GSM to UMTS
FDD_Qoffset
When to Reselect
Award Solutions Proprietary
The UE will perform a cell reselection from GSM to a UMTS reselection: 0 = - infinity (always select a cell if
neighbor cell if the following three criteria are met for a acceptable), 1 = -28 dB, 2 = -24 dB, … , 15 = 28 dB.
period of 5 seconds: Default value = 0 dB.
1. The RSCP of the measured UMTS cell exceeds that of Excerpt from 3GPP TS 45.008 regarding parameter
the GSM serving cell (and all suitable GSM neighbors) FDD_Qmin: A minimum threshold for Ec/No for UTRAN
by FDD_Qoffset, and FDD cell reselection, 0= -20dB, 1= -6dB, 2= -18dB, 3= -
8dB, 4= -16dB, 5= -10dB, 6= -14dB, 7= -12dB. Default
2. The UMTS cell’s measured Ec/No is greater than or
value= -12dB.
equal to the value (FDD_Qmin - FDD_Qmin_Offset),
and Excerpt from 3GPP TS 45.008 regarding parameter
FDD_Qmin_Qoffset: Applies an offset to FDD_Qmin value,
3. The UMTS cell’s measured RSCP value is equal to or
0 = 0 dB, 1 = 2 dB, 2 = 4 dB, 3 = 6 dB, 4 = 8 dB, 5 = 10
greater than FDD_RSCP_threshold, where
dB, 6 = 12 dB, 7 = 14 dB. Default value = 0 dB.
FDD_RSCP_threshold is equal to:
Excerpt from 3GPP TS 45.008 regarding parameter
• FDD_RSCPmin – min((P_MAX – 21 dBm), 3 dB) if
FDD_RSCPmin: A minimum threshold of RSCP for UTRAN
FDD_RSCPmin is broadcast on the serving cell,
FDD cell reselection, 0 = -114 dBm, 1 = -112 dBm, 2 = -
otherwise
110 dBm, 3 = -108 dBm, 4 = -106 dBm, 5 = -104 dBm, 6
• Qrxlevmin + Pcompensation + 10 dB, if these = -102 dBm, 7 = -100 dBm, 8 = -98 dBm, 9 = -96 dBm,
parameters are available, otherwise 10 = -94 dBm, 11 = -92 dBm, 12 = -90 dBm, 13 = -88
• the default value of FDD_RSCPmin dBm, 14 = -86 dBm, 15 = -84 dBm. Default value = -102
dBm.
Excerpt from 3GPP TS 45.008 regarding parameter
FDD_Qoffset: Applies an offset to RLA_C for cell
97
5 | Inter-Frequency and Inter-RAT Performance
Review Questions
1. Name the parameters used by the UE to calculate the
threshold to begin inter-frequency neighbor measurement.
2. What are the parameters used by the UE to compute
serving cell rank?
3. What could be the impact on idle mode UE behavior if
Qhysts parameter is set to high value?
4. What are the parameters/measurements used by the UE
to reselect GSM cell?
5. State True or False. Qoffset is broadcast in SIB 11/12 and
may have a unique value for each neighbor cell.
98
5 | Inter-Frequency and Inter-RAT Performance
99
5 | Inter-Frequency and Inter-RAT Performance
Connected Mode - I
IRAT/IF Handover Stages
Monitor/Trigger
Measurements
Define NL
Measurements
Report
Measurements
Handover
Execution
In addition to optimizing idle-mode reselection, the next 3. Once the UE detects a candidate that meets the
phase of ensuring an optimally performing network is to minimum reporting criteria, this is reported to the
understand the procedures and methods used while in network.
the connected mode. Handover is used in cellular mobile
4. The network evaluates the measurement results from
systems to maintain connections as UEs move between
the UE, and confirms that resources are available in
the coverage areas of different cells. To facilitate the
the target cell. If this evaluation is successful, the
handover process, the network uses the UE to monitor the
target cell is prepared to receive the hard handover
signal strengths of both serving cells and potential
and the handover command is sent to the UE. At this
handover candidates.
point, the UE executes the hard handover.
Inter-frequency and inter-RAT handover typically involves
four stages:
1. The handover process begins with a monitoring stage
where the mobile measures the current signal
strength of the serving cell. If the signal strength of
the serving cell drops below a predefined threshold,
the UE will inform the network.
2. At this point, the UE is provided a list of either inter-
frequency or inter-RAT target cells with the
appropriate reporting criteria, and is configured to
make out-of-band measurements.
100
5 | Inter-Frequency and Inter-RAT Performance
Connected Mode - II
UMTS DPCH Mobility Challenge
• COMPRESSED MODE (CM) – the mechanism by which the UE and the
UTRAN negotiate the time that must be taken away from data being
transmitted and received over the DPCH to permit tuning to a different
frequency or radio access technology for signal measurements
The UMTS DPCH has a number of challenges it must UMTS versus GSM with inter-frequency or inter-RAT
overcome in order to support mobility management, measurements is that (as the figure shows) the Dedicated
including: Physical Channel (DPCH) design does not support any
reserved bandwidth in the uplink or downlink to support
1. Since each neighboring cell in a UMTS system is
the time required to make these measurements. As a
typically operating on the same frequency as the
result, either the UE must be designed with two
serving cell, (or cells, in the case of soft handover),
independent receivers to support these measurements, or
the UE must perform intra-frequency measurements
a mechanism is required that enables a single-receiver UE
while still sending and receiving data on the Physical
to momentarily leave its serving carrier and take
Layer RF interface.
measurements.
2. As UMTS networks mature, additional carriers are
UMTS has adopted Compressed Mode (CM) as the means
needed to address capacity demands in high traffic
by which a more cost-effective single-receiver UMTS UE
areas. However, these new carriers may not be
could support the need for inter-frequency and IRAT
deployed everywhere, and may not be in the same
measurements. The compressed mode procedures are
frequency band, since not all areas require the
coordinated at the Physical Layer by the Node B and UE,
additional capacity or have the bandwidth for multiple
and are controlled via RRC signaling.
carriers.
3. UMTS must continue to support its legacy technology,
GSM.
In all of these cases, the UTRAN must be able to direct the
UE to tune to a different UMTS frequency or transition to
another Radio Access Technology (RAT). The challenge for
101
5 | Inter-Frequency and Inter-RAT Performance
10 ms frame
•Increased transmit power during this period to keep quality (FER, BLER,
BER…) unaffected due to reduced processing gain
Transmission More Power
Increased
gap (a few slots)
Power
Uplink
10 ms frame
Award Solutions Proprietary
102
5 | Inter-Frequency and Inter-RAT Performance
Uplink and downlink compressed mode operations can be 2. Another option is to simply adjust the rate of
configured separately and independently of each other; information from the upper layer, as long as the QoS
however, most implementations enable uplink and constraints imposed by the connection are met. This
downlink compressed mode at the same time. The is referred to as Higher Layer Scheduling (HLS) and is
measurement gaps are always set to occur at the same preferred for non-delay-sensitive services, such as
time in both directions. interactive packet-switched data.
There are two methods of implementing compressed The choice of which scheme to use depends on the radio
mode, to allow time for measurements without network load, the UE’s capabilities, the spreading factor in
interrupting the active voice or data session: use, and the type of service.
1. The first method reduces the spreading factor from
SF to SF/2, thereby reducing the time needed to send
data in frame by 50 percent. Since fewer small SF
codes are defined, this method may not be usable in
every case; for example, if the UE is already using an
SF 8 code, an SF 4 code may not be available for
compressed mode operation.
103
5 | Inter-Frequency and Inter-RAT Performance
104
5 | Inter-Frequency and Inter-RAT Performance
(Hyst/2)
event 1e or 2f
event 1f or 2d
(Hyst/2)
Time
H: Hysteresis
T: Time-to-trigger event 1f or 2d Triggered event 1e or 2f triggered
There are several reporting events that can be used to In addition, events 6a, 6b and 6d may also be used to
start and stop compressed mode measurements, manage the compressed mode. These are triggered when
depending on the specific implementation of the RNC. In the UE transmit power exceeds a threshold, falls below a
particular, events 1f and 2d indicate that the current cell threshold, or reaches maximum power, respectively.
has fallen below a minimum acceptable level, while 1e
Which specific event reports are used to trigger
and 2f indicate that the cell has recovered. The Threshold,
compressed mode for a UE will depend on the vendors’
Hysteresis and Time-To-Trigger (T) parameter
implementation of their hard handover algorithms.
configurations are delivered to the UE via Measurement
However, it is important to understand that all 3GPP-
Control messages. The actual measurements may be
compliant devices must be capable of using any of these
based on CPICH RSCP, CPICH Ec/No, or both, depending
event reports in a UMTS network.
on the information in the Measurement Control message.
If the measurement drops below the (Threshold –
Hysteresis/2) level for the specified T interval, a
Measurement Report indicating that the corresponding
event (1f or 2d) has occurred will be sent to the RNC. This
will typically result in compressed mode being triggered.
If the measurement becomes equal to or better than
(Threshold + Hysteresis/2) for T, a (1e or 2f)
Measurement Report will be sent to the RNC resulting in
the compressed mode operation being canceled.
105
5 | Inter-Frequency and Inter-RAT Performance
UE Node B RNC CN
Includes UE capabilities
RRC Connection Setup Complete (e.g., IRAT, freq, CM, etc.)
Measurement Control e1f or e2d (Ec/No) thresholds
The high-level call flow of a typical inter-frequency The event 2b thresholds define the minimum signal
handover is illustrated here. The first step in IFHO is the strength for both the serving cell and target neighbor cell
communication of the UE’s capabilities to the RNC. This required to initiate an IFHO.
includes, among other things, supported frequencies and
At this point, the UE begins evaluating IFHO neighbor cells.
the UE’s compressed-mode capability.
If a neighbor candidate meets the target cell event 2b
Assuming that the UE is capable of handling IFHO (or criteria, and the current cell is below the serving cell event
IRAT), it will be sent the event 2d and 2f parameters via a 2b criteria, the UE will send a Measurement Report to the
Measurement Control message. The UE uses these RNC indicating that event 2b has occurred.
parameters as reporting criteria while monitoring the
Upon receipt of an event 2d Measurement Report, the
signal strength of the serving UMTS cell. If the signal
RNC will confirm resource availability in the target cell and
strength of the serving cell drops below a threshold
configure a DCH in the target cell for the hard handover. At
(defined by the e1f or e2d parameters), a Measurement
this point, the RNC sends the UE a Physical Channel
Report indicating the event will be sent to the RNC.
Reconfiguration message instructing it to handover to the
After receiving the Measurement Report, the RNC will inter-frequency neighbor. The UE confirms a successful
send the UE a Physical Channel Reconfiguration message hard handover with a Physical Channel Reconfiguration
with compressed-mode-related parameters. After the UE Complete message.
has confirmed this message, it will be sent a
Measurement Control message containing a list of the
inter-frequency neighbors, the event 2b thresholds, and
instructions on when to activate the compressed mode.
106
5 | Inter-Frequency and Inter-RAT Performance
Thresh Used
Frequency
(H2b/2)
(H2b/2)
Thresh Non-Used
Frequency
H: Hysteresis Time
event 2b Triggered
T: Time-to-trigger
Award Solutions Proprietary
If the UMTS cell is configured to support IFHO, the If both cells simultaneously meet this criteria for the
Measurement Control message following the CM Physical defined Time-to-Trigger, a Measurement Report is sent to
Channel Reconfiguration procedure may contain the IFHO the RNC indicating an event 2b has occurred, identifying
Neighbor candidates (UARFCN, SC, etc.), and the event 2b the candidate cell that satisfied the non-used frequency
parameters. The event 2b parameters include Threshold event criteria.
Used Frequency (current UMTS cell), Threshold Non-Used
Frequency (target UMTS cell), Hysteresis, and Time-to-
Trigger (T).
Event 2b is triggered when the two following conditions
occur, for the defined T.
1. The UMTS serving cell’s Measurement Quantity
(CPICH RSCP or CPICH Ec/No) drops below the
threshold defined by (Threshold Used Frequency –
Hysteresis/2), and
2. the IFHO candidate pilot is greater than the threshold
defined by (Threshold Non-Used Frequency +
Hysteresis/2).
107
5 | Inter-Frequency and Inter-RAT Performance
UE Node B RNC CN
Includes UE capabilities
RRC Connection Setup Complete (e.g., IRAT, freq, CM, etc.)
Measurement Control e1f or e2d (Ec/No) thresholds
108
5 | Inter-Frequency and Inter-RAT Performance
UE Node B RNC CN
Includes UE capabilities (e.g.,
RRC Connection Setup Complete IRAT, freq, CM, etc.)
Measurement Control e1f or e2d(Ec/No) thresholds
Same as IFHO
Measurement Control
e1f or e2d (RSCP) thresholds
UE signaled to handover to
Handover From UTRAN Command
GSM
Award Solutions Proprietary
The high-level call flow of a typical IRAT handover is shown handover. At this point, the RNC sends the UE a Handover
here. The initial steps of this process, up to the initial From UTRAN Command instructing it to handover to the
Physical Channel Reconfiguration for compressed mode, GSM neighbor. The RNC confirms a successful hard
are the same as the IFHO case. In both cases, event 1f or handover via CN signaling between the 2G and 3G
2d is used to trigger compressed mode. networks.
After the UE has confirmed the Physical Channel
Reconfiguration message, it will be sent a Measurement
Control message containing a list of the IRAT neighbors,
the event 3a thresholds, and instructions on when to
activate compressed mode. The event 3a thresholds
define the minimum signal strength for both the serving
cell and the target GSM cell required to initiate an IRAT.
At this point, the UE begins evaluating IRAT neighbor cells.
If a neighbor candidate meets the target cell event 3a
criteria, and the current cell is below the serving cell event
3a criteria, the UE will send a Measurement Report to the
RNC indicating that event 3a has occurred.
Upon receipt of an event 3a Measurement Report, the
RNC will initiate Core Network (CN) signaling that will
confirm resource availability in the target GSM cell and
configure a traffic channel in the target cell for the hard
109
5 | Inter-Frequency and Inter-RAT Performance
UMTS event 3a
(H3a/2)
GSM event 3a
H: Hysteresis Time
event 3a Triggered
T: Time-to-trigger
A UMTS cell will typically be configured to support IRAT or Assuming the BSIC verification of the GSM candidate is
IFHO. Once a UE has reported an event 1f or 2d, and the successful, a Measurement Report message is sent to the
radio bearer has reconfigured to support the compressed RNC indicating an event 3a has occurred, identifying the
mode, a Measurement Control message is sent to the UE. candidate GSM neighbor that has satisfied that part of the
It is at this point that an IRAT handover will differ from an event criteria.
inter-frequency handover. This example shows the IRAT
case.
The Measurement Control message contains the GSM
neighbor candidates (ARFCN, BCCH, and BSIC) and the
event 3a parameters. The event 3a parameters consist of
Threshold Own System (UMTS), Threshold Other System
(GSM), Hysteresis, and Time-to-Trigger (T).
Event 3a is triggered when the two following conditions
occur, for the defined T:
1. The UMTS serving cell’s Measurement Quantity
(CPICH RSCP or CPICH Ec/No) drops below the
threshold defined by (UMTS 3a Threshold –
Hysteresis/2), and
2. The GSM candidate BCCH is greater than the GSM 3a
Threshold.
110
5 | Inter-Frequency and Inter-RAT Performance
UE Node B RNC CN
Includes UE capabilities (e.g.,
RRC Connection Setup Complete IRAT, freq, CM, etc.)
Measurement Control e1f or e2d (Ec/No) thresholds
Same as IFHO
Measurement Control
e1f or e2d (RSCP) thresholds
Radio Bearer Setup CM TGPS config. at RB setup
111
5 | Inter-Frequency and Inter-RAT Performance
112
5 | Inter-Frequency and Inter-RAT Performance
Summary
• The key drivers for IFHO and IRAT mobility include:
• Coverage (ensuring access to services throughout the
network) and
• Load Sharing (balancing traffic across the available
resources).
• Well-tuned reselection parameters are required to
ensure the UE is always camping on the correct cell.
• Hard handover triggers must strike a balance between:
– Preventing premature IRAT and IFHO (which impacts
system capacity and throughput), and
– Excessively delaying IRAT and IFHO (which may result in
dropped calls).
113
5 | Inter-Frequency and Inter-RAT Performance
Review Questions
1. Compressed mode procedure is used when the UE is in
_____________ mode.
2. The purpose of using the compressed mode procedure is
to____________________.
3. Which of the parameters are used by the UE to trigger
event 2d (start compressed mode measurement)?
4. What could be the impact on network performance if the
‘Time to Trigger’ value is set to high to delay IRAT and
IFHO?
114
6 | R99 Data and HSDPA Performance
Chapter 6:
R99 Data and HSDPA
Performance
115
6 | R99 Data and HSDPA Performance
Objectives
After completing this module, you will be able to:
• Recall the key aspects of R99 PS data and
HSDPA that enable it to deliver improved
downlink throughput
• Describe the channel switching mechanism for
resource optimization
• Understand the different approaches to
downlink power and OVSF code management
• Explain how CQI is derived, and its uses in
assessing and optimizing HSDPA performance
• Outline key factors that influence HSDPA
throughput
Award Solutions Proprietary
116
6 | R99 Data and HSDPA Performance
Packet Call
Establishment
117
6 | R99 Data and HSDPA Performance
3G-SGSN
UTRAN
5. PDP Context Activation
UE
Information Exchange
Here is a summary of the messaging required to set up a and Security mechanisms needed to ensure a secure
packet session in UMTS. This flow is conceptually similar communication. Authentication is optional while
to the circuit-switched flow; that is one of the fundamental security is mandatory.
advantages of a common Iu interface to the core network
5 In this step, the mobile supplies the packet session
and also a single RRC connection. Although the transport
information. In GPRS as well in UMTS, unless
layer functions and application packets change, it is
ciphering is established, none of the call related
essentially the same flow as the CS call. Detailed steps
information is sent over the air. This also sends the
are outlined in subsequent slides.
QoS requested by the mobile.
The first step in ANY messaging or service is the
6a-b In this step, several key stages of the call occur. The
establishment of an RRC connection. This is needed for
Radio Access Bearer Assignment message contains
the first service (CS or PS) that is invoking a service. There
all the Quality of Service (QoS) specific information
is only one RRC connection regardless of the domains
needed for the UTRAN to complete the packet
where the UE is connected . In this flow, the mobile
session. The UTRAN works with the Node B and the
requests an RRC connection to start a packet session.
mobile to ensure that this service is provided. Once
2-3 Once the RRC connection has been established, a completed, a positive response is sent that indicates
Service request for the packet session is sent. This is success in establishing a Radio Access Bearer, which
sent as a Direct Transfer message (which is routed is a combination of Radio Bearer and Iu-PS bearer.
through the SRNC) to the 3G-SGSN. This also
The packet session continues to completion after this
establishes the initial Iu-PS connection to the 3G-
point.
SGSN.
These summarized steps are expanded to illustrate the
4 This step is a summarized view of the Authentication
details involved in call establishment.
118
6 | R99 Data and HSDPA Performance
119
6 | R99 Data and HSDPA Performance
PS-Specific Metrics
User
Throughput
Delay
Throughput Cell
Efficiency Throughput
120
6 | R99 Data and HSDPA Performance
Throughput Factors
Available Network
codes resources
Available
power Traffic
levels
RAB rate
Application
Data
behavior
Packet data throughput can be limited by a wide variety of like Transmission Control Protocol (TCP), which are
factors. First and foremost, the RAB configuration sets the designed to improve end-to-end packet reliability, will slow
upper bound for throughput over the air: a 64 kbps RAB down the application when faced with transmission errors
will not be able to support a 100 kbps service. over the air interface. In these cases, the user will
experience a relatively low average throughput, even
Since the resources (codes and power) available on the
though there is nothing wrong with the network.
downlink are shared among all active users (both voice
and data), the user may be prevented from attaining a
higher data rate, due to code or power blocking. In
addition, the UE itself may not have enough power to drive
a higher speed link, especially as it moves away from the
cell.
Other factors may also adversely affect the maximum data
rate a user can achieve. Network limitations (such as Iub
capacity) may throttle the rate at which packets arrive at
the cell, while peaks of voice or data traffic may lead to
resource blocking.
Finally, the effects of the application itself should not be
forgotten. Many common applications (particularly those
involving Wireless Application Protocol, or WAP, servers)
simply do not generate very much data, even if the radio
link could sustain a higher rate. Also, certain IP protocols
121
6 | R99 Data and HSDPA Performance
122
6 | R99 Data and HSDPA Performance
Throughput Efficiency
Throughput efficiency can be thought of as the
percentage that the RAB “pipe” is filled.
Wasted capacity
Efficient usage
123
6 | R99 Data and HSDPA Performance
Low throughput
efficiency High traffic
levels
High traffic
Low throughput levels
efficiency
124
6 | R99 Data and HSDPA Performance
RRC States
RRC_Connected
URA_PCH Cell_PCH
Cell_DCH Cell_FACH
RRC_Idle
Radio Resource Control (RRC) is responsible for managing The UE will transition between the states as required,
all UEs being served by a particular RNC. From RRC’s ensuring that it uses only those resources appropriate to
perspective, a UE is either in the RRC_Idle state, in which its current activity. A UE can send or receive data only
case the RNC has no idea where the UE is and has no when it is in the Cell_DCH or Cell_FACH state.
means of talking to it, or it is in the RRC_Connected state,
and a signaling path exists between the RNC and the UE.
When the UE is first powered on, it starts in the RRC_Idle
state, and attempts to locate a network where it can
receive service. Once it finds a suitable cell, it registers
with the network, and enters the RRC_Connected state.
If the UE is not involved in an active voice or data session,
then the network merely tracks its location for paging
purposes, either to a UMTS Routing Area (URA) level, or
down to a specific cell. In this case, the UE is said to be in
the URA_PCH or CELL_PCH state, respectively.
If the UE starts a voice or data call, it is typically assigned
dedicated radio resources to support the service, and
moves to the CELL_DCH state. Under certain
circumstances, such as when the service does not require
a very high data rate, the UE may be moved to a low-
speed shared channel, in the CELL_FACH state.
125
6 | R99 Data and HSDPA Performance
Review Questions
1. What are the components of the Radio Access Bearer
(RAB) in UMTS?
2. List the key PS specific metrics.
3. List the factors which limit packet data throughput in
UMTS.
4. What is throughput efficiency?
5. How does dynamic RAB switching help with better
throughput efficiency?
126
6 | R99 Data and HSDPA Performance
HSDPA Operations
127
6 | R99 Data and HSDPA Performance
3
Key
WCDMA
Parameters
128
6 | R99 Data and HSDPA Performance
HS-SCCH
HS-DPCCH
HS-PDSCH
Node B Node B
UL DPCH UL DPCH
DL DPCH DL DPCH
HSDPA is an evolution of UMTS R99. Hence, HSDPA condition information on a continuous basis from all
supports all the UMTS R99 configurations without any the mobiles vying for access to the HS-DSCH. Each UE
restrictions, and also introduces new channels specifically will measure and determine the C/I value of each
designed to support high-speed packet data services. active set pilot and report the C/I of the best sector.
These channels are: Since HSDPA systems support the Hybrid ARQ (HARQ)
scheme, the transmitter (Node B) transmits some of
• High Speed – Downlink Shared Channel (HS-DSCH):
the turbo-encoded symbols first and waits for a
HS-DSCH is a new channel designed to carry high-
Physical Layer acknowledgement from the receiver
speed packet data traffic. Each cell may support one
(UE). If the response is a NACK, the base station
or more HS-DSCHs. The HS-DSCH channel is shared
continues to send additional symbols. If the response
across all users requesting HSDPA-specific high-
is an ACK, the base station stops sending the
speed packet data services. Sharing of the HS-DSCH
remaining symbols and continues with the next
is accomplished by using a combination of Time-
packet. The mobile sends these ACK/NACK
Division Multiplexing (TDM) and Code-Division
commands along with the current CQI on the HS-
Multiplexing (CDM) across multiple users.
DPCCH.
• High Speed - Shared Control Channel (HS-SCCH): HS-
Unlike R99 channels, the new HSDPA channels do not
SCCH is the control channel associated with the HS-
support soft handover. The HSDPA-specific channels are
DSCH. HS-SCCH conveys the HS-DSCH allocation
set up on a single “Serving HS Cell.” If the UE measures a
information, including the user identity, the number of
better candidate that meets the criteria defined by the
spreading factors used, and the modulation scheme
UTRAN (Measurement Control Message), a Measurement
to the UE.
Report will be sent and HS channels will be reconfigured
• High Speed - Dedicated Physical Control Channel (HS- on a new serving HS cell.
DPCCH): The HSDPA system gathers current radio
129
6 | R99 Data and HSDPA Performance
UE 1 10101010101010101
Node B
UE 1 4 ACK/NAK (HS-DPCCH)
Award Solutions Proprietary
Let’s discuss how the HSDPA High-Speed Downlink Hence, the scheduling algorithm and the assignment
Shared Channel (HS-DSCH) operates at a 10,000 foot approach will differ from one implementation to
level. Multiple UEs may be assigned to the HS-DSCH by another but will typically follow a proportional fair
the RNC since it is a shared channel. HS-DSCH operations share or round robin algorithm approach.
is summarized in the following four-step procedure:
3. The third step is the data transmission to the
1. The first step in the HS-DSCH operation sequence is selected user. When the scheduler selects a user, it
the Channel Quality Indicator (CQI) reporting by all the uses the CQI value reported by the UE and the data
UEs assigned to the HS-DSCH. Each UE on the HS- buffer waiting for transmission to decide the data rate
DSCH measures its radio conditions and provides the and modulation scheme for the transmission. The
Node B with an accurate idea of the current receiving Node B uses the selected configuration to send the
condition. The Node B gathers the CQI reports from data over the air.
all the UEs before proceeding to the next step. The
4. The UE receives the data and verifies the checksum.
UEs may report the CQI value once every 2
If the transmission was received properly, the UE will
milliseconds.
transmit an ACK to the Node B. If the transmitted
2. The second step is the execution of the scheduler at information was received with errors, the UE will send
the Node B to determine which user’s data should be an NACK to the Node B. This step involves the HARQ
transmitted next. The standards do not specify how aspect of HSDPA.
the scheduling algorithm must be implemented so
These steps are repeated continuously to support the HS-
this will vary between manufacturers.
DSCH effectively.
130
6 | R99 Data and HSDPA Performance
The HSDPA UE is a very complicated piece of equipment. • Minimum inter-TTI Interval - This is the maximum
The HSDPA specification has been designed to support reception frequency of 2 ms frames the UE can
various capabilities or categories of UEs. In the initial support. For example, a category 1 UE has a
release of HSDPA, it is understandable that the minimum inter-TTI interval of 3. This means that the
capabilities of the UE are limited to allow UE developers UE can only receive a transmission of data (i.e. a
an opportunity to learn from any initial deployment issues. transport block) from the Node B every 3rd 2 ms
The UE must be capable of communicating its capabilities frame. Whereas a category 6 UE, with an inter-TTI
to the UTRAN so the Node B will know how much data the interval of 1, can receive data every 2 ms frame. The
UE is capable of supporting. 3GPP specifications refer to lower the minimum inter-TTI interval, the higher the
this as the “Category” of the UE. A UE’s category defines potential data rate.
the maximum number of HS-DSCHs supported, the
• Maximum number of bits of an HS-DSCH transport
minimum number of TTI intervals a UE must wait before
block received within an HS-DSCH TTI” - This is the
receiving another data transmission from a cell, the
maximum transport block size that can be
maximum transport block size the UE can support and if it
transmitted to a UE, in a 2 ms frame. The higher the
can support 16QAM or QPSK only.
maximum transport block size the higher the
Here are the definitions of the capabilities listed in the maximum data rate the UE can support.
3GPP chart outlining these for the various UE’s HS-DSCH
• Total number of soft channel bits - This is the raw
categories:
buffer size that the UE must have to be able to
• Maximum number of HS-DSCH codes received - The support Hybrid ARQ (i.e. buffer incoming
first criteria specifies the maximum number of transmissions and retransmissions).
simultaneous 16 bit OVSF codes the UE can support
on the HS-DSCH.
131
6 | R99 Data and HSDPA Performance
132
6 | R99 Data and HSDPA Performance
133
6 | R99 Data and HSDPA Performance
DOWNLINK
R99 Voice and Data users
Available Power
Overhead Channels
Time
Award Solutions Proprietary
The Node B is provisioned to support a certain transmit needs to work with the downlink power management
power, say 20 to 40 Watts. A portion of this power is then philosophy of R99 and efficiently utilize the remaining
allocated to the control and overhead channels such as power to support high-speed data services. So, HSDPA
the Pilot, Forward Access Channel (FACH), Paging Channel utilizes any remaining HPA power for the TTI of 2 ms then
(PCH), etc. The remaining power is then dynamically adjusts modulation and coding based on the available
allocated among all the active users to support the HPA power for the next TTI period as shown in the figure.
forward traffic channels. Users at the cell edge typically By doing this, the HSDPA system is able to co-exist with
require higher power allocations due to the inferior RF multiple R99 voice users while providing high-speed data
channel conditions being experienced. Where users near services by utilizing the leftover HPA power.
the cell with a line of site to the antennas typically require
In other words, HSDPA adopts the philosophy of allocating
a much smaller power allocation to support their forward
the remaining HPA power to a user one TTI (2 ms) at a
traffic channels. Since voice traffic roughly uses a
time, and then adjusting the data rate based on the
constant data rate, link adaptation is more effectively
reported channel conditions and available HPA power for
achieved by adjusting the transmit power. This is also
the next TTI.
referred to as “fixed rate variable power,” where a cell
varies the transmit power while maintaining a fixed
throughput to the user regardless of where the user is
located in the cells coverage area.
As depicted in the figure, the available transmit power is
not fully utilized at all times in an R99 system. And, since
HSDPA systems are designed to support both voice and
high-speed data users simultaneously, the HSDPA traffic
134
6 | R99 Data and HSDPA Performance
Power Management:
Reserved Power for HSDPA
Max DL Power
Reserved (e.g. soft handover)
HPA Power
Time
Award Solutions Proprietary
Downlink power management becomes an even greater managed without any changes. However, since this
concern in HSDPA. The new downlink channels, the High method reserves a certain amount of the total HPA power
Speed – Downlink Shared Channel (HS-DSCH) and the for HSDPA, this power is no longer available for R99
associated High Speed Shared Control Channel (HS-SCCH) channels and services when there are no users on the HS-
are not fast power controlled like the previous traffic DSCH. This means that during peak traffic periods for
channels. Moreover, since the HS-DSCH and HS-SCCH are voice there is a risk of users being blocked from service
shared among users, the power applied to these channels due to a lack of downlink power resources in the Node B,
should be sufficient to reach the mobiles even at the which is typically not an acceptable option to operators.
periphery of the cells. The HSDPA specifications do not Similarly, since the amount of power allocated for the HS-
address this issue and leave it as an implementation DSCH is fixed in this scenario, any unused power
choice. As a result, the downlink power management resources available from the R99 pool cannot be taken
varies from one manufacturer to another leading to advantage of by HSDPA users.
product differentiators. There are many ways of
implementing the forward power distribution. Three of the
more common approaches are discussed on the following
slides.
The first approach deals with the HS-DSCH and HS-SCCH
power similar to the downlink overhead channels in UMTS
R99 configuration. The HS-DSCH and HS-SCCH channels
are assigned a reserved portion of the total HPA power of
the Node B cell. This method is very simple to implement
and permits the UMTS R99 channels and services to be
135
6 | R99 Data and HSDPA Performance
SF = 1 SF = 2 SF = 4 SF = 8 SF = 32 …
SF =16
In a UMTS R99 system, the downlink channels are length of the Spreading Factor codes could be quite short
identified by a unique Orthogonal Variable Spreading (down to 4 bits) on Dedicated Physical Channels (DPCH).
Factor (OVSF) code as shown in the Spreading Factor tree Thus, the co-existence of voice users and packet data
illustration. The Spreading Factor codes vary in length users, both based on DPCH and HSDPA packet data
from 4 bits to 512 bits. The Spreading Factor codes are channels, poses important challenges in OVSF code
generated as shown. Out of these Spreading Factor codes, management.
certain codes are reserved for overhead channels such as
the Pilot channel or the channel that broadcasts the
System Information Blocks for the cell. The total number
of OVSF codes at a given SF is the same as the SF. For
example, there are eight OVSF codes at SF=8. The HSDPA
system uses a fixed OVSF of 16. There are a maximum of
16 OVSF codes at SF=16. At the most, only 15 OVSF
codes at SF=16 are available for use with HSDPA.
Theoretically, these 15 codes can be assigned to a single
user during a TTI, or they can be shared among multiple
users during the TTI. The voice and HSDPA data users
share the same OVSF code tree.
Since HSDPA supports both voice and high-speed packet
data users, OVSF code management becomes very
critical. Spreading factors for voice users are typically of
128 bits in length. However, for R99 data users, the
136
6 | R99 Data and HSDPA Performance
3GPP specifications show that HSDPA is capable of You can see how fast the table starts to fill up. So if
supporting up to 15 SF=16 OVSF codes, because the first HSDPA is capable of supporting up to 15 SF16 codes, this
16 bit OVSF code C16,0 is reserved for R99 common may not be possible if the number of users in the HSDPA
channels. pool is high. Also if there are other R99 users, then even
fewer SF 16 codes will be available for HSDPA
Now lets add the HSDPA channels needed to support
applications. Actually for every 8 R99 voice users, an SF
multiple HSDPA users on the assumption that these users
16 code is used up.
are only doing data services (no voice). Two new downlink
channels are needed to be support one HSDPA user in the
downlink. The first is the common HS-SCCH (High Speed
– Shared Control Channel) which requires a 128 bit OVSF
code to be assigned. The second is a dedicated channel
(DPCH) per UE for carrying the RRC signaling since HSDPA
cannot carry signaling in Rel 5 configuration. For this
DPCH a 256 bit OVSF code is required. HS-SCCH is a
shared channel but a DPCH is required for every UE in the
HSDPA pool. Additional HS-SCCH channels will also be
required to support TDM configuration i.e. supporting
multiple UEs per TTI – one shared HS-SCCH required for
every UE per TTI.
137
6 | R99 Data and HSDPA Performance
CQI
138
6 | R99 Data and HSDPA Performance
139
6 | R99 Data and HSDPA Performance
CQI Fundamentals
Target BLER CQI Mapping:
≤ 10% TFRC Report
CQI
CQI Mapping Table: Consists of:
CQI value TFRC (TB Size, Modulation,
# of HS-PDSCHs)
∆: Ref Power : Measurement
Adjustment Power Offset
The Channel Quality Indicator (CQI) is calculated by the calculation process yields higher order CQI values
UE, based on an assumed power of the HS-PDSCH (PHS- that correspond to the maximum transport block size
PDSCH) and the reception quality of the CPICH. This of a given UE category. CQI values that map to
assumed power is derived by the UE using the following negative ∆ values simply indicated that a given
equation: transport block size can be decoded with even less
PHS-PDSCH
PHS-PDSCH = PCPICH + + ∆, where:
The CQI is reported to the serving HS cell. The Node B
1. is the Measurement Power offset sent in the Radio
uses the CQI mapping table (see next chart) as well as
Bearer Setup message indicating a positive or
other inputs (e.g. resource availability, buffer occupancy)
negative offset that should be applied to the
to determine the maximum size transport block, number
measured CPICH. In this manner the operator has
of codes, and type of modulation that will result in the
some parametric control over the UE’s assumption of
transport block being successfully decoded with no more
PHS-PDSCH. In future vendor implementations, gamma
than a 10% Block Error Rate (BLER). Because the Node
could be dynamically modified based on average HS
B’s available power for the HS-PDSCH is typically very
power availability in the Node B. The range of
dynamic, the UE can only assume what the PHS-PDSCH will
parameter is (-6 ... 13 dB, in 0.5 dB steps).
be at a given TTI.
2. ∆ is the Reference Power Adjustment, also called the
power reduction factor, which is mapped in the CQI
tables (TS 25.214) based on UE category and
reported CQI value. The purpose of this variable is to
de-rate the assumed PHS-PDSCH when calculating a CQI
value. This becomes necessary when the CQI
140
6 | R99 Data and HSDPA Performance
Using a vendor-proprietary algorithm, the UE estimates the dB more power at its disposal and determine that a 3300
DL RF channel conditions or CQI and reports this to the bit transport block can be sent using five codes and QPSK.
Node B, where it is cross-referenced with the CQI mapping Alternatively, the scheduler may only have one 16-bit
table to determine the TFRC or the largest transport block OVSF code available so it decides that a 400-bit transport
the UE could decode (based on the specified number of block can be sent using one code and QPSK.
OVSF codes and modulation type) at a less than 10-
It is important to note that the CQI report is only as reliable
percent BLER. However, it is important to note that this
as the chipset vendor’s RF receiver design and the
information is only used as a reference by the Node B
algorithm used to estimate the CQI value. This is why
scheduler. The scheduler not only knows CQI reported by
some network vendors implement additional features that
the UE, it also knows the actual available PHS-PDSCH. Based
can effectively offset the reported CQI. For example, if a
on this knowledge, the current code availability and the
UE is found to consistently over-estimate its CQI value,
other HS user requirements, the scheduler now has the
resulting in a BLER higher than 10 percent, then the
flexibility to schedule the actual transport block larger, or
scheduler may use this offset to correct future CQI reports
smaller than the size corresponding to the received CQI.
it receives from this UE type so data transmitted to the UE
For example, the UE, based on the current PCPICH is capable of being decoded consistently at the expected
measurement, determines the assumed PHS-PDSCH, then BLER or better.
uses this to estimate its CQI to be 12. The Node B then
receives this CQI and, according to the CQI mapping table
for a category 7 or 8 UE, determines the UE should be
able to decode a 1742 bit transport block sent on the HS-
PDSCH using three OVSF codes and QPSK, at better than
a 10-percent BLER. However, the scheduler may have 3
141
6 | R99 Data and HSDPA Performance
142
6 | R99 Data and HSDPA Performance
Review Questions
1. How many HS-SCCHs are needs to be configured to
schedule two users in 2ms TTI ?
2. In HSDPA, how does the reporting of the CQI (Channel
Quality Indicator) help with achieving higher data rates?
3. How is HSDPA different from R99 packet data operation?
4. State True or False.
1. In HSDPA, the power for the transmission of the High
Speed Downlink Shared Channel is fixed.
2. There is no Soft Handover support in HSDPA.
143
6 | R99 Data and HSDPA Performance
HSDPA Performance
144
6 | R99 Data and HSDPA Performance
RF Coverage Optimization
Good RF coverage control is essential for optimal
performance
• Site location, antenna type, height, azimuth and downtilt have a
greater impact on HSDPA
• Ensure a dominant server exists for good pilot pollution control
All cellular technologies perform at their best with careful An additional unwanted side-effect of poorly-optimized
RF coverage control. This is particularly true for CDMA- neighbor lists is the possibility of entering the Compressed
based technologies like WCDMA, and even more true for Mode (CM) in the core of the HSDPA service area as a
“C/I-critical” technologies such as HSDPA. Careful result of reaching the CM Ec/No or RSCP trigger
selection of cell site parameters – site location, antenna conditions.
height, antenna type, antenna azimuth, and antenna
downtilt – result in marked improvements in throughput.
Specifically, server dominance (a strong “C”) should be
emphasized, as well as the minimization of pilot pollution
(a strong “I”). Pilot pollution is commonly defined as four
or more servers at a given location, all of which are within
a few dB (typically around 5 dB) of each other in strength.
Whereas WCDMA systems, which utilize soft handoff, can
be somewhat forgiving of pilot pollution, single-server
technologies like HSDPA cannot reach their full
throughput potential in the presence of rampant pilot
pollution.
Neighbor list optimization goes hand-in-hand with
coverage optimization. Having clear server dominance and
little pilot pollution is useless unless the UE is informed of
the correct neighboring servers through optimized
neighbor lists.
145
6 | R99 Data and HSDPA Performance
Channel Reliability
Channel reliability impacts throughput
146
6 | R99 Data and HSDPA Performance
There are numerous UTRAN parameters related to HSDPA DPCCH, the less power there is for the other uplink
throughput performance, though most of them are “set- channels, and the closer the UE must be to the cell site.
and-forget” in nature. Once these parameters are
Instead of increasing the amount of power applied to the
optimized, they will likely not be re-optimized on an
CQI and ACK/NACK messages, repetition can be used. A
ongoing basis. Further, the majority of the cell sites in a
message transmitted twice requires (ostensibly) half the
network will use identical values for these parameter
amount of power per transmission; however, there are
settings.
drawbacks. In the case of CQI, as the repetition factor is
e1a and e1b govern the typical Active Set size, and increased, the UE is less able to respond to rapidly
HSDPA UEs will use the same parameters as R99 UEs. changing channel conditions. This, in turn, could impact
There is some flexibility in setting e1d which, again, throughput in two ways - through a loss of throughput
causes the UE to issue a measurement report indicating a opportunity (in case of rapidly improving conditions), and
change in the best server. Since e1d includes a through a loss of actual throughput due to HS-DSCH errors
hysteresis, higher values make the UE more resistant to (in cases of rapidly degrading conditions). Repetition of
relocating its serving HS-DSCH, thus reducing the amount the ACK/NACK messages causes a proportionate
of HS-DSCH ping pong. At the same time, throughput will reduction in throughput since the cell has to wait longer to
be impacted, since higher values also result in the UE receive the ACK/NACK before deciding to send a new
being served by a less-than-optimal cell more of the time. packet or to retransmit the current packet.
The CQI, ACK and NACK channel gains and repetition With most deployments, the HS-DSCH is permitted to
factors govern the reliability of the HS-DPCCH. A highly- consume most or all of the remaining power and code
reliable HS-DPCCH is critical to maximizing throughput, but channels after the R99 users/channels are served.
the drawback to over-powering the HS-DPCCH is reduced Obviously, any restriction on the power and code
uplink coverage. The more power applied to the HS- resources available to HSDPA will reduce throughput.
147
6 | R99 Data and HSDPA Performance
Cell Throughput
Cell loading (R99 traffic) limits HSDPA
peak throughput performance per cell
•Balance of RF power and code channel sharing
148
6 | R99 Data and HSDPA Performance
Summary
• Best effort is the approach taken for non-real-time R99
data; nevertheless, by channel/rate switching, we can
optimize resources available for other service types (e.g.,
voice/HSPA).
• HSDPA performance relies on the leftover power availability
at the Node B.
• CQI, BLER and cell change frequency are some important
HSDPA metrics that can be used for optimization.
• Handsets are complicated. Numerous UE categories limit
HSDPA data rates, potentially impacting subscriber
expectations.
• Backhaul is very expensive and HSDPA throughput is easily
limited by Iub capacity.
• When possible, HSDPA should be the preferred mode for
data transmission regardless of required data rate.
Award Solutions Proprietary
149
6 | R99 Data and HSDPA Performance
Review Questions
1. How does incorrect neighbor list configuration affect
HSDPA throughput performance?
2. List the system parameters which affect downlink
throughput performance.
3. How does poor coverage affect HSDPA throughput?
4. How does the best server PSC plot help in analyzing
HSDPA throughput performance?
150
7 | HSUPA and HSPA+ Performance
Chapter 7:
HSUPA and HSPA+
Performance
151
7 | HSUPA and HSPA+ Performance
Objectives
After completing this module, you will be able to:
• Explain the key aspects of HSUPA that enable it to
deliver improved uplink throughput
• Understand the code assignments for HSUPA
• Identify the parameters utilized in the diagnosis and
optimization of HSUPA performance
• Describe how HSPA+ improves performance in terms of
data rates, capacity and battery life
• Explain new UE categories and their characteristics
defined in HSPA+
152
7 | HSUPA and HSPA+ Performance
Uplink Interference
Management
153
7 | HSUPA and HSPA+ Performance
UL Interference Management
Load
RoT Threshold
Thermal Noise
Each UE adds to the noise level. The noise level must be Time
managed so that it stays below a threshold.
When a UE transmits, it adds to the total uplink support high speed packet data services in the uplink.
interference. Unlike the downlink, the uplink cannot rely Uplink interference management is therefore the most
on orthoganality between users to reduce interference. important factor to consider when optimizing HSUPA to
This means that UEs transmitting on dedicated channels provide enhanced data performance in the uplink. HSUPA
in the serving cell and all of the users transmitting on the has been designed in such a way that it has the ability to
E-DCH channels are contributing to the overall manage the available power margin very efficiently so as
interference levels. In addition, the UEs currently to maximize the data throughput supported in the uplink.
transmitting information on channels in any of the This is also accomplished without sacrificing channel
surrounding cells also contribute to the interference levels quality to existing users already assigned dedicated
seen by the Node B. resources, i.e., R99 voice & data users.
As a result, an acceptable threshold limit needs to be set;
this threshold will also define the capacity or throughput
the cell is capable of supporting. This limit is called the
Rise over Thermal (RoT) threshold. It represents the
amount of interference that can be added to the uplink
frequency above the thermal noise level before the
interference becomes too high to decode any uplink
information.
As was seen in HSDPA in the downlink, the goal of HSUPA
is to share the remaining power margin after accounting
for the various contributing sources to noise levels, and
use this to allocate dedicated channels (E-DCH) to
154
7 | HSUPA and HSPA+ Performance
Thermal Noise
Each UE adds to the noise level. The noise level must be Time
managed so that it stays below a threshold.
HSUPA uses fast power control and soft handover to can simply make one common grant to all users during
control reverse link interference levels, very similar to the absolute grant process and then use relative grants as
R99. In order to enhance throughput and reduce latency, a means of fine tuning resource assignments to each
HSUPA also introduces a new uplink scheduling capability users current needs. Conversely, 3GPP supports the ability
in the Node B, as well as support for HARQ in the uplink. to make grants that can be tailored by QoS requirements,
Similar to the downlink scheduler HSDPA introduced in the both between users (inter-user QoS) and within the same
Node B, the uplink scheduler is responsible for device (intra-user QoS).
determining the code and power requirements a UE will
Since service-based QoS is not widely used today by
be given when an uplink high speed packet data service
wireless broadband service providers, it is unlikely that
request is received. Unlike R99, where the Node B was
any of the current vendor implementations will use the
just a relay between the RNC and UE, the Node B now has
inter- or intra-user QoS approach. Instead, the focus here
added intelligence to support the enhanced data
will be on the configuration and optimization of the
throughput requirements needed by both HSUPA and
absolute and relative grant procedures, in order to
HSDPA.
maximize data throughput per cell in the uplink, as well as
The HSUPA uplink scheduler algorithm is not defined by maintaining a good user experience.
the 3GPP specifications; it is left to the vendors to decide
how this function is implemented in their network
infrastructure. The uplink scheduler is a proprietary
performance differentiator between network vendors.
What the standard does provide is flexibility to make to
make uplink resource allocation as simple or complex as
the vendor chooses or the operator demands. The Node B
155
7 | HSUPA and HSPA+ Performance
HSUPA Operations
156
7 | HSUPA and HSPA+ Performance
While HSUPA and HSDPA both deal with high-speed can do to best use its resources. Every mobile
packet access, there are some fundamental differences reports its current channel receiving conditions
between the two. for its best cell (also referred to as the serving
cell). Only the best cell for a UE will schedule and
1. Soft Handover:
transmit data to it. Other nearby cells may be
a. HSDPA tries to use the available power impacted; however, the impact does not affect
effectively by allocating it for high-speed DL transmissions, since the reported CQI is
downlink data transmissions. These decisions based on current conditions.
are made every 2 milliseconds. Since the
b. In HSUPA, the best cell (a.k.a., the serving cell)
available power varies from cell to cell, HSDPA
will allocate serving grants, indicating how much
does not use soft handover.
power the UE can use for uplink E-DCH
b. HSUPA supports soft handover for uplink transmissions. The serving cell makes its
transmissions. When a UE transmits, energy is decision solely based on its measured uplink
sent in all directions. In other words, if a UE is interference levels; however, the UE’s
transmitting energy, it will impact the UL transmission impacts all other cells in its vicinity.
interference levels in all the cells in its vicinity. Hence, HSUPA maintains the active set for a UE
Therefore, it is better to keep the UE in soft on the E-DCH and allows the non-serving cells in
handover and retain the ability to control the UE its active set to be able to adjust the total uplink
transmit power from all nearby cells. power used by a UE. In other words, non-serving
2. Scheduling: cells can send an “SOS” to the UE and tell it to
reduce its power (and therefore uplink
a. HSDPA scheduling is based on what each cell interference).
157
7 | HSUPA and HSPA+ Performance
158
7 | HSUPA and HSPA+ Performance
HSUPA-Related Channels
Noise Control
Scheduling
Non-serving E-AGCH/E-RGCH
Cell C1 Scheduling/Noise Control
E-DPDCH / E-DPCCH
UE
Noise High-Speed Data and Signaling
Control Serving
E-HICH Cell C3
ACK/NACK
This diagram captures the role of different channels 4. When the E-DCH active set cells receive data, the
involved in HSUPA. In this example, the Enhanced Hybrid ARQ process at each cell transmits an ACK or
Dedicated Channel (E-DCH) has an active set consisting of a NACK on the E-DCH HARQ Indicator Channel (E-
three cells from three different Node Bs. The serving E- HICH), indicating the success of the packet
DCH cell is identified by the RNC as part of the radio link transmission. Note that the E-HICH and E-RGCH are
setup process through RRC signaling; cells that are not assigned the same channelization code. The E-
part of the serving cell’s Node B are referred to as the RGCH/E-HICH can also be shared by many UEs in a
non-serving E-DCH radio link set. given cell.
The HSUPA channels are processed by the UE as follows:
1. The UE receives Absolute Grant information from the
E-DCH serving cell on the E-DCH Absolute Grant
Channel (E-AGCH). The E-AGCH may be shared by
multiple UEs in a given cell.
2. The UE also receives Relative Grants (i.e., corrections)
from the serving and non-serving E-DCH radio link set
on the E-DCH Relative Grant Channel (E-RGCH).
3. The UE transmits the E-DCH transport channel data
on the uplink using the E-DCH Dedicated Physical
Data Channel (E-DPDCH), and associated control
information is sent on the E-DCH Dedicated Physical
Control Channel (E-DPCCH).
159
7 | HSUPA and HSPA+ Performance
HSUPA Transmission
1 Scheduling Request
Run the Scheduling
Algorithm
3 Granting allowed resources
Scheduler
High-Speed Data +
4
Associated signaling Node B
5 Ack/Nack
3
4
5
...
Award Solutions Proprietary
The HSUPA process can be described in five basic steps. transmits an ACK to the UE. If the transmitted
Note that the duration of time between the first and last information was received with errors, the Node B
step is on the order of 10 ms or so. sends a NACK to the UE.
1. The UE makes a scheduling request. These steps are repeated continuously to support the
high-speed data process effectively and efficiently.
2. The serving Node B scheduler receives the scheduling
requests from all of the UEs it managed. The
scheduler’s primary function is to determine the
amount of power to allocate to each UE; the
standards do not specify a specific scheduling
algorithm.
3. Resources are granted to the selected users. When
the scheduler selects a set of users, it uses the uplink
channel quality and the data bandwidth requests
from the UE to decide the resource allocation. The
Node B signals the resource allowances over the air
to the UEs.
4. The UE sends the data to the Node B, along with the
associated control information.
5. The Node B verifies the checksum of the data. If the
transmission was received properly, the Node B
160
7 | HSUPA and HSPA+ Performance
161
7 | HSUPA and HSPA+ Performance
HSUPA UE Categories
Maximum Number Minimum Maximum
UE HSUPA Supported E-
of Codes Code Data
Category DCH TTIs
Transmitted Size Rate
Category 1 1 SF4 0.7 10 ms only
162
7 | HSUPA and HSPA+ Performance
HARQ Processing at UE
C1 C3
Serving
UE1 Cell
If one cell sends an ACK,
the packet is successful
Nack
C2
Award Solutions Proprietary
Hybrid ARQ processing in HSUPA is fundamentally A single ACK is enough for the UE to consider the
different from the HSDPA case, since the transmission is corresponding transmission as a success and move on
one-to-many in HSUPA as opposed to one-to-one in with the transmission of the remaining subpackets. From
HSDPA. This introduces a number of challenges to the the network’s point of view, the cells that did not positively
network and the UE in high-speed uplink operations. acknowledge the packet initially expect a retransmission,
since they do not know whether the packet was received
During a call, the UE is constantly updated with the list of
correctly by the other members of the E-DCH active set.
cells that it will be communicating with for the E-DCH. This
However, the MAC-e layer at the Node B is indirectly
list defines the E-DCH active set for the UE, which may be
informed about the status of the previously transmitted
a subset of the normal active set of the R99 traffic. After
subpacket, since the UE indicates the transmission of a
the mobile transmits its payload in a TTI, it knows when
new subpacket by resetting the retransmission sequence
and how to expect the feedback information for that
number (RSN) to zero. Now, the Node Bs that had earlier
particular packet since the HARQ process is strictly
“NACKed” the previous packet know that a new subpacket
synchronous in HSUPA.
is being received, which allows them to reset their own
Since the information is received by the cells in the active MAC-e buffers.
set at the same time (within the propagation delay time,
which is negligible), each cell transmits its feedback
information (ACK or NACK) simultaneously, based on
whether the received block was decoded successfully. At
this point, the UE must decode these messages on the
corresponding E-HICH channels of each cell and
determine their value.
163
7 | HSUPA and HSPA+ Performance
The minimum requirement for a cell that is configured to UE is required for signaling. In Release 7, an option to
support HSUPA is an E-RGCH, an E-HICH, and an E-AGCH. support signaling on HSDPA is available. If this is
implemented, then DPCCH per UE would be required for
The E-RGCH and E-HICH share one 128-bit code, and the
power control; however, with the need to conserve codes,
E-RGCH/E-HICH will themselves be shared by multiple
a new Fractional Dedicated Physical Channel has been
users, through the use of signature sequences (up to 20
defined in R7 to support power control for up to 10 UEs.
users per code). Thus multiple E-RGCH and E-HICH
channels may be required based on expected maximum
HSUPA capacity
In addition, one E-AGCH must be deployed in the cell,
which requires a 256-bit code. This channel is also shared
by all of the UEs in the cell, but signature sequences are
not used; instead, the assigned E-RNTIs are used to
differentiate grants to different UEs.
164
7 | HSUPA and HSPA+ Performance
Review Questions
1. Which of the following capabilities /features is NOT utilized
in HSUPA?
1. Power Control
2. Soft Handover
3. Adaptive Modulation
4. Variable spreading factor
2. List the sources of interference within the uplink of a cell
in a high-speed system.
3. Name the channel that the UE uses to send high-speed
data to the Node B in HSUPA.
4. What is meant by ‘Grant’ in HSUPA?
5. In HSUPA, which of the approach is adopted by the Serving
Cell and Non-Serving Cells to manage Grants for a UE?
Award Solutions Proprietary
165
7 | HSUPA and HSPA+ Performance
HSUPA Performance
166
7 | HSUPA and HSPA+ Performance
167
7 | HSUPA and HSPA+ Performance
168
7 | HSUPA and HSPA+ Performance
Review Questions
1. In the HSUPA serving PSC plot, if the UE only receives
“hold” indications from the surrounding cells in a
significant portion of time, what do you understand by
this?
2. In Hybrid ARQ NACK plot, if the NACK rate is 80 percent in
significant portion of time, what does this indicate?
3. How is the UE itself a key limiting factor for low uplink
throughput?
169
7 | HSUPA and HSPA+ Performance
170
7 | HSUPA and HSPA+ Performance
Release 9
Release 7 Release 8
Dual Band Dual
Cell HSDPA
Higher Order Dual cell HSDPA
Modulation (HOM) Dual Cell HSDPA Dual Cell HSUPA
64 QAM (DL),
Dual Cell HSDPA
16 QAM (UL)
MIMO with 64 with MIMO
QAM
MIMO
Improved Layer 2 Release 10
Improved Layer 2 (UL)
(DL) 4 Carrier HSDPA
Improved Layer 2 (DL)
The objective of HSPA+ (3GPP R7, R8 and R9) is to helps to enhance peak throughput to 42 Mbps.
enhance HSPA performance in terms of spectrum
Improved Layer 2 (DL and UL): 3GPP R7 and R8 have
efficiency, peak data rate and latency. Key features that
done certain modification to Layer 2 (RLC/MAC) to
help to enhance throughput are briefly described below.
support higher data rates. From 3GPP R7 and R8, RLC
64 QAM in DL and 16 QAM in UL.: With 64 QAM, the supports flexible RLC PDU size in both directions instead
network pushes more data bits to the user and therefore of a fixed size like legacy HSPA. New MAC entities MAC-
increases the downlink data rate with no change in ehs (R7) and MACi/is (R8) are introduced to support
bandwidth. In HSPA+, rate control of the high speed flexible RLC PDU size.
channel is done primarily through changing the
Dual cell HSDPA and MIMO: 3GPP Release 8 introduced a
modulation and/or coding scheme. 3GPP R7 allows the
new feature called Dual Cell HSDPA to enhance user peak
UE to use 16 QAM on E-DCH (Enhanced Dedicated
throughput up to 42Mbps in downlink. Dual cell operation
Channel) in the uplink and therefore HSPA+ R7 can
is characterized as simultaneous reception of two High
achieve uplink peak data rates of 11.5 Mbps.
Speed – Downlink Share Channel (HS-DSCH) transport
MIMO and 64 QAM: Multiple Input Multiple Output channels on two different carriers. Two carriers must be
(MIMO) means the use of multiple antennas at the adjacent and should belong to same Node B. However two
transmitter and receiver side. Independent data streams, cells should not use MIMO to serve UEs configured for
using the same time-frequency resource, are sent over dual cell operation. 3GPP Release 9 allows the use of
different antennas. This provides multiplexing gain and MIMO with dual cell HSDPA and hence helps to increase
increases system capacity and data rates. MIMO allows peak throughput up to 84 Mbps.
downlink peak data rate of 28 Mbps with 16 QAM.
However, 3GPP R7 does not support MIMO with 64 QAM.
3GPP R8 allows support of 64 QAM with MIMO and hence
171
7 | HSUPA and HSPA+ Performance
Release 9
Release 7 Release 8
Dual Band Dual
Cell HSDPA
Higher Order Dual cell HSDPA
Modulation (HOM) Dual cell HSDPA Dual Cell HSUPA
64 QAM (DL),
Dual Cell HSDPA
16 QAM (UL)
MIMO with 64 with MIMO
QAM
MIMO
Improved Layer 2 Release 10
Improved Layer 2 (UL)
(DL) 4 Carrier HSDPA
Improved Layer 2 (DL)
Dual Band Dual Cell HSDPA: Dual Band Dual Cell feature
in 3GPP HSPA+ R9 enhancement allows simultaneous
reception of two HS-DSCH transport channel which are
carried over different frequencies from different frequency
bands. However these two carriers should be from the
same Node B.
Dual Cell HSUPA: In 3GPP Release 9, a new feature called
Dual Cell HSUPA or Dual Cell E-DCH was introduced to
enhance user uplink peak throughput to a maximum of 23
Mbps. In Dual Cell E-DCH operation, the UE is configured
with two uplink frequencies and two downlink frequencies.
For Dual Cell HSUPA, the UE must be configured with Dual
Cell HSDPA on a single frequency band.
Four Carrier HSDPA: In 3GPP Release 10, several features
have been combined and expanded to utilize four carriers
in the downlink. These can be in separate bands or in one.
The carriers in the same band are adjacent, or contiguous.
When four carriers, 64 QAM modulation, and 2 x 2 MIMO
are all utilized, peak data speeds of 168 Mbps can be
achieved on the downlink.
172
7 | HSUPA and HSPA+ Performance
173
7 | HSUPA and HSPA+ Performance
R7
e c a – MOD
Space-Time f ed c b a
f ed c b a
Receiver
f d b – MOD
174
7 | HSUPA and HSPA+ Performance
RNC
Flexible RLC
PDU size UE Flexible RLC
HSPA+
PDU size
PDCP Node B RRC
RLC MAC-ehs RLC
MAC-ehs MAC-i MAC-is
MAC-i/is PHY
PHY
• MAC–ehs can perform • MAC–is does macro
• MAC–i/MAC-is can MAC segmentation diversity selection in
perform MAC • Supports flexible RLC case of soft handover
segmentation PDU size and performs MAC
• Supports flexible RLC • MAC–i issues reassembly
PDU size scheduling grants
Prior to HSPA+, RLC protocol supported only fixed RLC • Since the RLC packet size is increased, the number
PDU size. Therefore, the use of MIMO, HOM or dual cell in of packets to be processed is clearly reduced and
HSPA+ will help to increase peak data rate only at the unnecessary segmentation operations also can be
Physical Layer. However, the peak data rate at the RLC avoided.
Layer is limited by the fixed RLC PDU size.
In HSPA+, three new MAC entities MC-ehs and MAC-i/-is
Layer 2 enhancements (includes both RLC and MAC) are defined.
become necessary to support the high data rates. To
MAC-ehs and MAC –i entities allow the support of flexible
enhance peak data rates even at the RLC level, the RLC
RLC PDU sizes as well as MAC segmentation/reassembly.
has been enhanced to support a flexible RLC PDU size.
The MAC-ehs at Node B in downlink and MAC-i at the UE
When the flexible RLC PDU size is configured by higher
in uplink perform the necessary segmentation of MAC-ehs
layers, the data PDU size is selected according to the
and MAC-i SDUs respectively. In addition, these two
payload size. Flexible RLC PDU size is only supported
entities handle multiple streams from Dual Cell and/or
when mapped on the HS-DSCH or E-DCH.
MIMO configurations and handle the common transport
The flexible RLC offers the following benefits: for Enhanced Cell_FACH.
• The relative RLC overhead is reduced due to Like MAC-es in HSUPA, MAC-is at the SRNC combines
increased packet size. The effective application data packets received from different Node Bs to perform
throughput is thereby increased due to reduced maximum ratio combining when the UE is in soft
overhead. handover. Also, as with MAC-e in HSUPA, MAC-i at Node B
issues scheduling grants to UEs.
• Large RLC block sizes lead to higher data rates.
175
7 | HSUPA and HSPA+ Performance
Dual Cell HSDPA (DC-HSDPA) 2. The two cells must operate in the same frequency
band.
3GPP Release 8 introduced a new feature called Dual Cell
HSDPA to enhance user peak throughput up to 42 Mbps. 3. The two cells have the option to use MIMO in Rel 9
Dual cell operation is characterized as simultaneous (see next chart).
reception of two HS-DSCH transport channels on two
Dual Band Dual Cell HSDPA (Dual Band DC-HSDPA)
different carriers. One of the carriers is called the primary
or anchor carrier at the serving cell and another carrier is DC-HSDPA was introduced in 3GPP Release 8 with
called the secondary or supplementary carrier in the restrictions that two carriers should be from the same
secondary serving cell. In the uplink, the UE shall use only frequency band and must be adjacent. This could be an
the primary carrier to send feedback information on single issue in areas where an operator has spectrum on two
HS-DPCCH for both HS-DSCH transport channels. different bands and they do not have 10 MHz available for
UMTS on either band. The Dual Band Dual Cell feature in
The dual cell HSDPA feature aids to fulfill the desire for
3GPP HSPA+ Rel 9 enhancement allows simultaneous
better and more consistent user throughput across the
reception of two HS-DSCH transport channels which are
cells in particular, when radio conditions are such that
carried over different frequencies from different frequency
existing techniques like MIMO cannot be used.
bands. However these two carriers should be from the
However, there are some restrictions when deploying this same Node B. There is no option for MIMO with Dual Band
feature. These restrictions are discussed below for Rel-8. DC-HSDPA.
1. Both serving and secondary servings cells should On the uplink, HSUPA operates using a single cell in Rel
belong to the same Node B and should be on 7/8. There is an option to use Dual Cell HSUPA in Rel 9
adjacent carriers. (see next chart).
176
7 | HSUPA and HSPA+ Performance
DC-HSDPA with MIMO Dual Cell E-DCH operation has the following
characteristics:
As seen in the earlier section, 3GPP R8 supports both the
dual cell HSDPA and MIMO features to enhance peak user • It supports only 2ms TTI.
throughput but with the restriction that MIMO cannot be
• In the uplink, the UE shall use each activated uplink
used for a dual cell configured UE. With 3GPP R9, a new
frequency for E-DPDCH, E-DPCCH and DPCCH
feature called Dual Cell HSDPA with MIMO was
transmission. HS-DPCCH is transmitted only on the
introduced. It allows a dual cell configured UE to work with
primary carrier.
MIMO. With this feature, theoretical peak user throughput
is increased to 84 Mbps. MIMO can only be configured • In the downlink, E-AGCH, E-RGCH and E-HICH are
with adjacent dual cell operation. In the uplink, only one transmitted on a corresponding frequency configured
channelization code is used for HS-DPCCH to carry to the uplink frequency.
feedback information ACK/NACK, CQI, PCI related to two
HS-DSCH transport channels.
Dual Cell HSUPA
In 3GPP release 9, a new feature called Dual Cell HSUPA
or Dual Cell E-DCH was introduced to enhance user uplink
peak throughput to a maximum of 23 Mbps. In Dual Cell
E-DCH operation, the UE is configured with two uplink
frequencies and two downlink frequencies. For dual cell
HSUPA, the UE must be configured with Dual Cell HSDPA
on a single frequency band.
177
7 | HSUPA and HSPA+ Performance
178
7 | HSUPA and HSPA+ Performance
Release 7 Release 8
Key features that help to enhance UE battery life and HSPA+ allows usage of HSPA resources in Cell_FACH,
reduce latency are briefly described below. URA_PCH and Cell_PCH state along with Cell_DCH state.
CPC – DTX: In HSPA R6 networks, the uplink DPCCH is Direct Tunnel Architecture: Another way to improve HSPA
transmitted continuously, even if there is no transmission performance is through a flatter architecture. The one
activity on the Enhanced Dedicated Channel (E-DCH) or tunnel solution allows the user plane to bypass the SGSN.
HS-DPCCH. This increases uplink channel overhead. In The advantage of the one tunnel solution is that there are
HSPA+, if there is no uplink data transmission on the E- fewer network elements (only two elements) in the user
DCH or HS-DPCCH, uplink DPCCH is not transmitted plane. It can efficiently support VoIP services at reduced
continuously. It is transmitted with a known activity latency.
pattern defined per UE.
CPC-DRX: Prior to HSPA+, the UE was required to monitor
the HS-SCCH continuously to decode HS-DSCH
information. This leads to increased HS-SCCH overhead
and UE battery consumption. In HSPA+, the network can
limit the number of subframes where the UE has to
monitor the HS-SCCH. It not only reduces the power
consumption, but also reduces interference in the cell and
consequently higher capacity can be achieved.
Enhanced Cell_FACH (DL and UL): In legacy HSPA
operation, HSPA resources are available only in the
cell_DCH state. Enhanced Cell_FACH state feature in
179
7 | HSUPA and HSPA+ Performance
180
7 | HSUPA and HSPA+ Performance
HSPA HSPA+
Cell_DCH
• Possible in
Possible to use
Not
Not possible
possible HSPA+ services
HSPA channels
to use
in HSPA
HSPA Cell_PCH/ • HSPA channel
channels
services URA_PCH enhancement
181
7 | HSUPA and HSPA+ Performance
UE Categories (Release 7)
HS-DSCH Maximum Minimum Highest Order
Maximum Support
UE Number of Inter-TTI Modulation
TBS (bits) for MIMO
Category OVSF Codes Interval Scheme
13 15 1 35280 64QAM No
14 15 1 42192 64QAM No
15 15 1 23370 16QAM Yes
16 15 1 27952 16QAM Yes
35280 64QAM No
17 15 1
23370 16QAM Yes
42192 64QAM No
18 15 1
27952 16QAM Yes
Recall that there are 12 HSDPA UE categories and 6 This is the highest data rate supported by HSPA+ in
HSUPA UE categories from the Physical Layer perspective. Release 7. For the UE category 18, when 64QAM (and not
A UE category is associated with capabilities of the UE’s (2x2) MIMO) is used, the peak data rate is 42192 bits/2
Physical Layer. HSPA+ increases the total number of ms= 21 Mbps. In cases of enhanced HSUPA, a category 7
HSDPA categories to 18 by introducing six new UE UE supports 16QAM for the UL E-DCH transmission,
categories. It adds one more HSUPA category, bringing the allowing a Transport Block of 22996 bits to be sent in a 2-
HSUPA categories to 7. The HSDPA UE category specifies ms TTI. This gives the highest HSPA+ rate of 22996 bits/2
the maximum number of OVSF codes (at SF=15) the UE ms = 11.5 Mbps in the uplink.
can process, the gap between two successive HS-DSCH
transmissions (1 meaning a gap of 0 TTIs), the maximum
size of one transport block, the highest-order modulation
scheme the UE can handle on the HS-DSCH, and an
indication of whether the UE supports (2x2) MIMO or not.
Release 7 HSPA+ either allows 64QAM or (2x2) MIMO but
not both at the same time for a given UE during a TTI of 2
ms. Consider the enhanced HSDPA category 18. When
(2x2) MIMO is supported, the maximum transport block
size is 27952 bits (which is the same as the case of
normal HSDPA UE category 10). Since a (2x2) MIMO
enables transmission of two such blocks, a total of 27952
*2 = 55904 Bits are transmitted in the 2-ms TTI. This
leads to the peak data rate of 55904 bits/2 ms = 28
Mbps.
182
7 | HSUPA and HSPA+ Performance
E-DCH UE Categories support minimum SF=2, minimum inter-TTI interval = 2 ms and maximum
of 4 codes for each E-DCH transport transmission
Category TBS (bits) Highest modulation scheme ( Dual Cell)
8 11484 QPSK
9 22996 16 QAM
TBS – Maximum no. of bits of HS-DSCH TB received per TTI
Dual Cell - 2 HS-DSCH i.e., 2 TBS / TTI ; Dual Cell +MIMO - 2 TBS/HS-
DSCH, i.e., 4 TBS / TTI Award Solutions Proprietary
183
7 | HSUPA and HSPA+ Performance
Summary
• HSUPA takes advantage of the unused interference budget on the
uplink to allow UEs to transmit at a higher data rate.
• Uplink throughput can be limited by many factors, including: UE
capabilities, total uplink traffic and interference levels, system
parameters, and network resource limitations (e.g., backhaul capacity).
• HSPA+ key features help to enhance throughput and reduce latency/UE
power consumption
DL UL
R7 64 QAM, MIMO (2x2), Improved Layer
2 , Enhanced Cell_ FACH, DTX/DRX
Direct Tunnel Architecture
R8 MIMO + 64 QAM, DC-HSDPA Layer 2 enhancements,
Enhanced Cell_FACH
R9 Dual Band Dual Cell, DC-HSDPA + DC-HSUPA
MIMO
.
Award Solutions Proprietary
184
7 | HSUPA and HSPA+ Performance
Review Questions
1. Name the key HSPA+ features that help to improve
throughput.
2. Name the key HSPA+ features that help to reduce latency.
3. State True of False. The eNB requires twice the power for
MIMO operation compared to without MIMO. What
benefits are obtained if CPC-DTX feature is enabled in the
network?
4. State True or False. Layer 2 enhancement helps to
improve application throughput.
185
186
Acronyms
187
Acronyms
dB Decibel
dBm Decibel per Milliwatt
DCCH Dedicated Control Channel
DCH Dedicated Channel
DL Downlink
DPCCH Dedicated Physical Control Channel
DPCH Dedicated Physical Channel
DPDCH Dedicated Physical Data Channel
DRX Discontinuous Reception
DTCH Dedicated Traffic Channel
DTX Discontinuous Transmission
E-AGCH E-DCH Absolute Grant Channel
Eb/No Energy per Bit over Noise
Eb/Nt Energy per Bit over Noise and Interference
Ec/Io Energy per Chip over Interface
Ec/No Energy per Chip over Noise
E-DCH Enhanced Dedicated Channel
EDGE Enhanced Data Rates for Global Evolution
E-DPCCH Enhanced Dedicated Physical Control Channel
E-DPDCH Enhanced Dedicated Physical Data Channel
E-HICH E-DCH Hybrid ARQ Acknowledgement Indicator Channel
E-RGCH E-DCH Relative Grant Channel
E-RNTI E-DCH Radio Network Temporary Identifier
E-TFCI E-DCH Transport Format Combination Indicator
FACH Forward Access Channel
FBI Feedback Information
FBI Final Block Indicator
FCC Federal Communications Commission
FDD Frequency Division Duplex
F-DPCH Fractional Dedicated Physical Channel
FER Frame Error Rate
FTP File Transfer Protocol
GANSS Galileo and Additional Navigation Satellite Systems
GGSN Gateway GPRS Support Node
GMM GPRS Mobility Management
GPRS General Packet Radio Service
GPS Global Positioning System
GSM Global System for Mobile Communications
HAP Hybrid ARQ Process
HARQ Hybrid ARQ
HCS Header Check Sum
188
Acronyms
189
Acronyms
190
Acronyms
191
Acronyms
192
References
Standards
1. 3GPP TS 23.009: “Handover Procedures”
2. 3GPP TS 23.018: “Basic Call Handling – Technical realisation”
3. 3GPP TS 23.022: “Functions related to Mobile Station (MS) in idle mode”
4. 3GPP TS 23.101: “General UMTS Architecture”
5. 3GPP TS 23.110: “UMTS Access Stratum; Services and Functions”
6. 3GPP TS 23.121: “Architecture Requirements for release 99”
7. 3GPP TS 23.920: “Evolution of GSM Platform towards UMTS”
8. 3GPP TS 23.925: “UMTS Core network based ATM transport”
9. 3GPP TS 23.930: “Iu Principles”
10. 3GPP TS 25.201: “Physical Layer – General Description”
11. 3GPP TS 25.211: “Physical channels and mapping of transport channels onto physical channels”
12. 3GPP TS 25.212: “Multiplexing and channel coding (FDD)”
13. 3GPP TS 25.213: “Spreading and Modulation (FDD)”
14. 3GPP TS 25.214: “FDD; physical layer procedures”
15. 3GPP TS 25.231: “Physical layer measurements”
16. 3GPP TS 25.301: “Radio Interface Protocol Architecture”
17. 3GPP TS 25.302: “Services provided by the physical layer ”
18. 3GPP TS 25.303: “UE functions and inter-layer procedures in connected mode”
19. 3GPP TS 25.304: “UE procedures in Idle Mode”
20. 3GPP TS 25.321: “Medium Access Control (MAC) Protocol Specification”
21. 3GPP TS 25.322: “Radio Link Control (RLC) Protocol Specification”
22. 3GPP TS 25.331: “Radio Resource Control (RRC) Protocol Specification”
23. 3GPP TS 25.401: “UTRAN Overall Description”
24. 3GPP TS 25.410: “UTRAN Iu Interface : General Aspects and Principles”
25. 3GPP TS 25.413: “UTRAN Iu Interface RANAP signaling”
26. 3GPP TS 25.420: “UTRAN Iur Interface: General Aspects and Principles”
27. 3GPP TS 25.430: “UTRAN Iub Interface: General Aspects and Principles”
28. 3GPP TS 25.433: “UTRAN Iub interface NBAP signaling”
29. 3GPP TS 25.434: “UTRAN Iub interface data transport & transport signaling for CCH data streams”
30. 3GPP TS 25.832: “Manifestations of Handover and SRNS relocation”
31. 3GPP TS 25.922: “RRM Strategies”
32. 3GPP TS 25.931: “UTRAN Functions, examples on signaling procedures”
33. 3GPP TS 25.832: “Manifestations of Handover and SRNS relocation”
193