The future potential of High Speed Uplink Packet Access in existing 3G networks

FRIDAY 24th AUGUST 2007 No. 10

The future potential of
High Speed Uplink Packet Access in
existing 3G networks

September 2007

An Omnitele White Paper

Turning opportunities into success

The future potential of High Speed Uplink Packet Access in existing 3G networks

The 3rd generation cellular mobile networks are continuously undergoing heavy development, and
especially technologies related to packet data transmission are evolving rapidly. In order to overcome
the issue of uplink being a bottleneck for data services an upgrade, WCDMA Enhanced Uplink (EUL), is
specified in the most recent version of UMTS standard release 6.
In this paper concrete results from a study assessing the upcoming performance of the new system is
presented. The main goal is to give an implication of how the Enhanced Uplink will perform in existing
3G networks and even more important, to show what the key factors are affecting achievable bitrates
and the quality of service.
Due to the complexity of the system and different available optimisation strategies, remarkable differ-
ence between vendors’ equipment is to be expected, at least in the beginning of EUL evolution. Dur-
ing the study it was also evident that although high bitrates of over 5 Mbps are promised for Enhanced
Uplink, only approximately 2.5 Mbps is achievable in practical large scale networks, even with the most
advanced future terminals. Still, according to the case study carried out, at least the denser part of the
investigated network was very suitable for EUL implementation. It was found that in most cases the ter-
minal capabilities are not the limiting factor and the site density is mostly sufficient for providing high
data rates of up to 1.4 Mbps in the early phase of the evolution.
From operator point of view, the system seems to be extremely feasible. Mainly with software updates
an operator is able to reach improved quality of service, capacity and coverage, hence we expect rapid
penetration.

Introduction
The demands for today’s cellular mobile networks early 90’s offered at maximum 9.6 kbps throughput

2 are growing continuously. The operators want to for a single user. Today the commercial state-of-
improve the spectral efficiency of their systems the-art High Speed Downlink Packet Access (HSDPA)
since spectrum resources are scarce and expen- transmission schemes specified for third generation
sive. At the same time the users want to have bet- WCDMA (Wideband Code Division Multiple Access)
ter QoS (quality of service) in terms of increased networks bring bitrates of several megabits per
bitrates and reduced latencies. Evidently this is second for a single user; the current defacto-speed
a difficult equation and a lot of research and de- seems to be 3.6 Mbps and even 7.2 Mbps is expected
velopment work is done all over the world to ful- in the near future. In short, in approximately fifteen
fil not only current, but future needs as well. years a thousand-fold performance improvement to
cellular mobile systems has been introduced when
While designing the future improvements for exist- considering data services.
ing mobile networks, several aspects should be kept As already mentioned, the most recent remarkable
in mind. The new features should degrade the per- feature implemented widely to commercial 3G net-
formance of the existing network as little as possi- works is the HSDPA architecture introduced in 3GPP
ble. On the other hand, to be deployable with rea- WCDMA standard release 5 (R5). Still, high downlink
sonable amount of work, the new features should bitrates e.g. from base station (BS) to user equip-
utilise the existing network infrastructure as exten- ment (UE) are not enough for the service portfolios
sively as possible. This leads us to a difficult equa- the operators want to offer since the uplink direc-
tion, with the fundamental question being: “How to tion e.g. from UE to BS quite soon becomes the bot-
pump even more resources out of a system already tleneck. To improve the R5 uplink throughput, a new
running close to its limits?” When getting closer to architecture called Enhanced Uplink (EUL) was pre-
the best achievable QoS, the value of even a one sented in December 2005 in 3GPP Standard for WCD-
decibel improvement can bring an operator or a net- MA Release 6. To underline that the EUL belongs to
work vendor crucial advantage against a competitor. the same “family” with HSDPA, it is often referred
–That is exactly the reason why we can see the new to as HSUPA – High Speed Uplink Packet Access, and
technical abbreviations day after day on marketing the “family” is often called simply HSPA – High Speed
brochures. Packet Access.
When looking back at the cellular mobile systems and When carrying out the study in June 2007, vendors
especially the data services, one can see that the were preparing to implement their pilots and EUL
development has been revolutionary. The Circuited capable user terminals were becoming available on
Switched Data (CSD) service that was introduced for the market. A wide range of simulations and theoret-
GSM (Global System for Mobile Communications) in ical analysis had been carried out by several organi-

© Omnitele Ltd. 2007


zations to forecast the impacts of the improvements The retransmitted frame is soft combined with the
introduced in EUL. It is commonly expected that the original frame, and it turns out that bit energy, e.g.
early phase implementations of EUL will improve the transmit power can be saved.
single user maximum physical layer UL throughput Besides Hybrid ARQ and NodeB Scheduling, the new
to 1.4 Mbps and to even 5,7 Mbps in the near future. standard introduces an optional feature of shorten-
Also from system efficiency point of view a signifi- ing the transmission time interval (TTI) to 2 milli-
cant performance improvement of even 70%-100% in seconds. The goal is to further improve the accuracy
increased cell throughput compared to R5 UL can be of scheduling by reducing the minimum resource al-
achieved in some circumstances. Since the perform- location block size, a good strategy though. How-
ance increase of EUL can be implemented mainly ever, the gains of this feature are expected to ap-
with software upgrades to R5 (e.g. HSDPA capable) pear mostly in higher bitrates (e.g. not near future
networks, rapid penetration can be expected. deployments) of HSUPA, hence it is defined to be
However, there is still a momentary lack of experi- optional. The feature is commonly referred simply
ence from real life implementations, and no one ac- as Short TTI.
tually knows how well the EUL will perform in exist- The fourth enhancement to the existing uplink ar-
ing 3G grids. This fact was the reason why Omnitele chitecture is Multicode Transmission, by which we
decided to carry out an extensive investigation un- mean a transmission scheme utilizing more than one
der the topic, including theoretical analysis, simula- spreading code simultaneously. In fact, the feature
tions and a special case study with field measure- was specified already in R99 uplink but not a single
ments, and to be able to respond to operator needs vendor has implemented it in real life products. The
and ever increasing consultative demands. goal of multicode transmission is simply improving
the dynamics of the system. In concurrent deploy-
ments, the maximum specified WCDMA R99 bitrate
HSUPA – Basic Principles of 384 kbps is achieved rather easily; hence excess
bandwidth can be achieved without using more re-
sources. According to our best knowledge, most of

3
Right after HSDPA found its way to commercial ap- the vendors’ equipment will in the beginning support
plications, the rumours from its “younger sister”, multicode transmission with two codes of spreading
HSUPA started to spread. The common belief was factor four, yielding maximum bit rate of 1.4 Mbps.
that very soon we would see an uplink version of The maximum specified bitrate with future termi-
HSDPA enabling high bit rates from mobiles to base nals shall be introduced by four codes, thus yielding
stations. However, such implications were probably 5.4 Mbps.
adopted due to poor and inaccurate marketing slide The four key features, their interaction and con-
ware, as the two technologies actually have very lit- tributions to the overall performance increase are
tle in common. So when thinking at the properties of summarized in figure 1. IT is worth noting is that
HSUPA one should forget about HSDPA, and instead none of the features themselves would alone intro-
consider it as a very fast and accurate UMTS R99 duce tremendous gains to the system, but they are
uplink. In fact, HSUPA is not a stand-alone feature designed to support each other in the best possible
at all and it relies highly on the normal WCDMA up- way. It is the short TTI that provides the other fea-
link. tures with the means to operate accurate enough;
it is the node B scheduling that enables the Hybrid
ARQ to function fast enough, and so on.
The Key features of HSUPA
Probably the most important improvement intro- Multicode
transmissio
HARQ Fast BS
Scheduling
Short TTI

duced by HSUPA is relocating the scheduler, e.g. the n

“operative brains” of the network, to the base sta- UE capability
beyond 384 kbps
Cell Capacity
Gain 20-50%
Coverage Gain
0.5 – 1 dB
Latency Gain:
RTT < 50ms
tion, instead of having it in the radio network con-
troller (RNC). By this way the scheduler can react
faster to the radio circumstances and hence distrib- HSUPA
uting the resources can be done significantly faster.
The latter is often referred to as NodeB Scheduling. Peak
datarate
Cell
throughput
More
coverage for
Better end
user QoS

Another new feature that further improves the net- >5 Mbps gain high bitrates

work coverage and capacity is Hybrid ARQ (Auto-
matic Repeat and Request). Whereas the traditional Figure 1. The picture illustrates the features of
ARQ discards an erroneously received packet and HSUPA and the respective effects on user level
requests a retransmission, the new Hybrid ARQ uti- and system level.
lises the information of erroneous frames as well.



HSUPA potential in a 3G net- RoT (before admission control etc. starts to operate)
work in metropolitan Helsin- in a real life network is always defined, the absolute
ki - case study maximum for the provided carrier to noise ratios
(CNR) is determined as well. Allowing a reasonable
load, a typical parameter value for highest allowed
To derive actual results that give implications of up- RoT is set to 6 decibels. Now, according to several
coming HSUPA performance in concurrent network, link simulations the highest HSUPA bitrates require
Omnitele consultants conducted an extensive study significantly higher CNR than what can be provided
on topic including theoretical analysis, simulations with RoT of 6 decibels. To be more specific, accord-
and field measurements in metropolitan area of Hel- ing to our analysis, bitrates higher than 2.5 Mbps
sinki. The complete target area, which was divided simply cannot be achieved in concurrent without in-
into two sections according to site density, is pre- creasing the site density. And moreover, the latter
sented in figure 2. holds for future terminal equipments as well.

Path loss limitations
As it has been already explained, knowing the
planned RoT limits of a certain network, the maxi-
mum achievable bitrates can be determined utilis-
ing link simulations and path loss measurements.
According to our measurements, at least in the ur-
ban parts of the target area, the site density was
sufficient for providing high bitrates of HSUPA. And
furthermore, it was found that the 1.4 Mbps, which
Figure 2. The target area of metropolitan Hel- is the expected maximum at starting point, could be

4
sinki was divided into Urban and Suburban/Rural achieved rather easily; in Helsinki city centre only
sections according to site density. The blue line ~10% of samples indicated worse throughput than
shows the drive test route. 1.4 Mbps to be achieved from the coverage point of
view. Some parts of rural/suburban target area still
To summarise the key findings, we found out that suffered from coverage issues yielding significantly
there are two factors that have a rather dominant lower HSUPA throughput. In figure 3 we represent
effect to the achievable HSUPA bitrates. Firstly, the the cumulative densities of achievable throughputs
distributed resource in WCDMA uplink is the pro- for an early phase implementation (Category 3 sup-
duced uplink noise rise, e.g. rise over thermal (RoT) ported) and a future deployment (Category 6 sup-
and secondly the uplink bitrates are naturally lim- ported) of HSUPA in the investigated area illustrated
ited by path loss, e.g. signal attenuation on radio in figure 2.
path. Those three parameters, RoT, path loss and
bit rate, form a static triangle that define the con-
Cum ulative Throughput Distribuition - Urban Area

figuration. No matter whether the case is building a 1
0,9
CAT3 Urban
CAT6 Urban
new HSUPA capable network or implementing it to 0,8
Cumulative Probability

CAT3 Rural
0,7
CAT6 Rural
an existing one, the dimensioning strategy focuses 0,6
0,5

on making a fair trade-off between the mentioned 0,4
0,3

variables. High RoT yields poor coverage but high 0,2
0,1
bit rates and good overall capacity utilisation. Low- 0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600

er RoT on the other hand leads to bigger coverage Thrpughput [kbps]

hence more severe path loss and lower bitrates. In
this sense the HSUPA works exactly like traditional Figure 3. Cumulative density functions of achiev-
R99 WCDMA uplink. able path loss limited HSUPA throughputs in the
investigated area.

Rise over thermal limitations
Now, coming back to the case study of metropolitan
Helsinki, it can be pointed out that two of the men-
tioned variables are fixed in any implemented 3G
network, hence determining the third should not be
an impossible task. Furthermore, as the maximum



Conclusions
In the latter section we saw that the estimated
packet call throughputs especially in the urban case
are rather near to the specified maximum data rate
of the utilized model. So, what’s left then to ana-
lyse? Isn’t it now so, that implementing the needed
software updates of EUL into the existing network
by a push of Enter-button brings a data rate of 1.4
Mbps to almost every user in the system? Well, not
quite. Every model has its weaknesses and as long as
experiments from large scale real life implementa-
tions are absent, no one can actually reliably say
how the future implementations will perform.
The right conclusion to make from the case study is
that when implementing the EUL into existing net-
works the achievable bitrates are not typically lim-
ited from the UE side. Hence at least the measured
network, and especially the urban part, is dense
enough to provide high EUL data rates. In the rural
parts of the target area the bitrates will not be so
high in general, as the UE capabilities will limit the
performance at least to some extent. And moreo-
ver, our modelling only investigated the maximum
achievable bitrates, with only one user in the sys-
tem; hence capacity issues were not covered.

5
During the study we were also able to point out the
most important factors related to dimensioning of
HSUPA networks. This is clearly utilisable and impor-
tant information for a network operator.
As Omnitele already has strong competence on the
field of HSDPA, we continue developing our HSUPA
service products as well, and by the end of year
2007 Omnitele will have a complete planning, audit-
ing, optimising and benchmarking product offering
for 3G HSPA - High Speed Packet Access solutions for
Operators world wide.

Omnitele Ltd.
Omnitele Ltd. is an internationally recognised telecommunica-
tions consultancy company providing comprehensive advisory
services for mobile operators in Europe, Africa, Middle East
and the Caribbean. The company was established in 1988 and
is owned by Finnish national telecom operator group Finnet. We
offer a wide range of Network and Business Consulting services
including mobile network planning and optimisation, network and
business development, and operator product optimization.

Publications details
Published September 2007
Author Mr. Mikko Valtonen, Consultant, Omnitele Ltd.

For more information, please contact:
Veronica Stjernvall, Marketing Manager, Omnitele Ltd.
firstname.lastname@omnitele.fi


Omnitele Ltd. Helsinki
Tallberginkatu 2A
P.O.Box 969
00101 Helsinki
Finland

Tel. +358 9 695991
Fax. +358 9 177182

contact@omnitele.fi

www.omnitele.fi

The future potential of High Speed Uplink Packet Access in existing 3G networks

More Related Content

What's hot (20)

Viewers also liked (9)

Similar to The future potential of High Speed Uplink Packet Access in existing 3G networks (20)

The future potential of High Speed Uplink Packet Access in existing 3G networks