|
|
 Why
do this work?
|
Contributing partners
Do you want to have a condensed version
of this information at your fingertips ? Here is a short PowerPoint
presentation (in pdf format) to download.
|
In
1992 the World Administration Radio Conference (WARC) identified a block
of spectrum between 1900MHz and 2200MHz which was designated for use by
Third Generation Mobile Systems. The idea was that these new systems would
satisfy the predicted growth in mobile traffic, and extend the capabilities
of fixed networks to mobile users. Now in 2000, the telecommunications
industry has developed a family of technologies for use in this spectrum,
and governments around the world are beginning to licence the spectrum
to operators of Third Generation mobile networks. The technology of choice
within Europe is based on the 3GPP
system UMTS, for which the first release of the specifications was recently
completed. The radio access network for UMTS is based on Wideband CDMA
(W-CDMA), which is significantly more complex than GSM, and as such introduces
a number of new challenges to radio network designers and planners.
The purpose of this activity on P921 has been to consider the UMTS Radio Access Network (UTRAN) from a design perspective, and provide guidelines on the design of a UTRAN.
| IMT-DS | Based on ETSI W-CDMA, using direct spread CDMA (Code Division Multiple Access) |
| IMT-MC | Based on US cdmaOne, using multi-carrier CDMA |
| IMT-TC | Based on ETSI TD/CDMA, using time and code division multiple access |
| IMT-SC | Single Carrier - based on UWC-136 / EDGE (Enhanced Data Rates For GSM And TDMA/136 Evolution) |
| IMT-FT | Frequency Time - based on DECT |
Learn more about UMTS & W-CDMA
Learn more about Applications and services possibilities (Review of foreseen UMTS applications) and QoS parameters (QoS measures for applications).
Impact of the UMTS Radio Interface on Mobile Multimedia services/applications
Concerning mobile multimedia applications, which have very demanding QoS requirements, further investigation have been performed. In fact, transport over radio interface can result errors and delay, so in order to keep and acceptable level of QoS, application layer mechanisms and UMTS radio control function are required.
In EURESCOM Project P921 framework three kind of applications have been selected for quality of service analysis. The objective of the quality test is to assess the impact of the UTRA (UMTS Terrestrial Radio Access) interface on the selected applications: audio retrieval and video download MPEG-4 applications and IP-based applications (web browsing, ftp). The test methodology followed was to use the error patterns obtained by means of a link level simulator to corrupt the application bit stream and to evaluate the degradation of the quality due to the radio interface.
The results of the tests have shown a strong impact of the UTRA interface on the quality of service. For example, real time streaming of high quality music over UMTS requires very ‘protected’ channel, at least when the application is not applying any error resilience tools. A UMTS network design has to take into account capacity, QoS and coverage requirements. If you are interested in more information about application testing follow the link to the application testing executive summary.
This dynamic behaviour makes cell planning and network dimensioning a very complex process. Traditional static prediction methods are not appropriate and so simulation and statistical modelling techniques have to be used. However, the system is very complex, with so many interactions, that the simulation has to be broken into two parts:
Learn more about link level simulation
ETSI selected W-CDMA and TD-CDMA as radio interfaces for the UMTS system. The main focus of the work performed in this task was to establish a link level simulator, which can analyse the radio propagation channel, taking into account UTRA's physical layer features. The link performance is a necessary input for system capacity and coverage evaluations that are definitely the most important aspects for a telecommunication operator.
Two link level simulators (W-CDMA and TD-CDMA) have been developed in the framework of the P921 EURESCOM to evaluate the radio performance of UTRA (UMTS Terrestrial Radio Access), in particular the majority of UTRA’s physical layer features have been simulated:
Link level simulators comprise the whole radio chain:
The
main outcome of link-level simulations is the minimum Eb/No
(ratio between energy per bit and noise) that is required by each service
in certain conditions (environment, mobile speed, coding/interleaving scheme)
to reach the threshold bit error rate (BER) or block error rate (BLER).
For speach with a required BER of e.g. 10-3, the network has
to provide a Eb/No of 5.7 dB. The figure provides
in addtion the BER available by hard or soft-coded Turbo codes (TBC). The
Eb/No value is one of the most important parameters
to design the radio access network, where Eb is mainly affected
by transmit power, antenna configuration and propagation environment, while
No is mainly affected by the traffic in the network.
The link-level simulations have been performed in four scenarios; two vehicular and two Indoor to outdoor, with different channel responses. Voice and data (circuit-switched and packed-oriented) services have been analysed, both in W-CDMA and TD-CDMA. The working point (minimum Eb/No required to reach the QoS requirements) has been determined for each set of configuration parameters. For example, in a vehicular environment the system working point for the voice service varies from 8.2 dB to 15.6 dB in the downlink and from 6.2 to 13.6 in the uplink for mobile user speed varying from 3 km/h to 500 km/h, as indicated in the following figure.
The link simulation results of the FDD
component, show for the voice service an independence of the working point
with respect to the mobile speed in the range 3-250 km/h. In case of data
service (LCD & UDD), the working point is more sensitive to the mobile
speed and to the propagation environment.
 |
|
|
Concerning the TDD component the link performance for the 8 kbit/s voice service and the 384 kbit/s data service have been analysed. The link performance of the TDD mode is more influenced by the mobile speed than the FDD mode. This behaviour is related to the round trip delay of the power control algorithm, which is higher in the TDD mode than in the FDD one. Moreover, the absence of the antenna diversity in downlink causes a sensitive degradation of the link performance, with respect to the uplink, especially for circuit switched data service.
A further output of the link level simulations
is a set of error patterns to be used for the test of a selected application.
If you are interested in more information
about link-level simulations follow the link to the link
level simulations executive summary
Learn more about system-level simulation
In UMTS capacity and coverage are interdependent and are functions of the radio environment. With an increasing number of users in the cell, the radius of the cell will decrease and the noise level, seen as interference, will increase (the attached figures gives an example for voice services in UMTS). Furthermore, both uplink and downlink radio performance have a large impact on system performance. QoS of the applications is also becoming a function of the location, the load of the network and the traffic mix. Multi-services and packet data transmission with packet dependent QoS parameters complicate the picture even more.
System-level simulations have the goal to model a 'complete' network with several base stations and an estimated traffic distribution with a mix of services, each one with different requirements in terms of error rates and delay. The values of Eb/No for each service that allows to accomplish with the maximum error rates are obtained by means of link level simulations. The main output of system level simulations is coverage, capacity and QoS in the simulated network.
A system-level simulator is basically a software tool modelling the system and the environment. Any mobile network system simulator is composed of the following blocks: base station, mobile user equipment, propagation model, data collector, graphic interface manager and simulator manager.
There are two approach to design system-level simulators: the time based and the snapshot. Both approaches provide complementary cellular network results, but due to the complexity of time based system-level simulations, most of the simulators that are currently available use statistical methods, known as Monte Carlo simulation.
The outcome of the system level simulator is the coverage area, as shown in the figure. Each cell is formed by one sector of the three sector antenna (this example). The shaded areas are the areas where coverage from two or more cells occure, the 'macrodiversity areas'. In UMTS terminals will have connections to both cells, to allow for fast and optimised handover at high data rates.
One of the main disadvantages of system-level simulators is the difficult of validating the results. To verify the conclusions obtained by means of simulation is very useful to compare results obtained by different institutions and organisations (universities, equipment suppliers, operators) using the same simulation scenarios. In the P921 project several scenarios and simulations guidelines have been specified.
If you are interested in more information about system-level simulations follow the link to the system-level simulations executive summary.
)).
UTRAN, the UMTS Terrestrial Radio Access Network, operates in two modes, the UTRA FDD and the UTRA TDD mode.
UTRA FDD (W-CDMA) specifications are being developed within the 3GPP. This radio access scheme is direct-sequence CDMA with information spread over approximately a 5 MHz bandwidth with a chip rate of 3.84 Mcps. The radio interface carries a wide range of services to support both circuit-switched services and packet-switched services. A radio protocol has been designed where several different services such as speech, data, multimedia can simultaneously be used and multiplexed onto a single carrier. Bearer services support real-time and non-real time applications by employing transparent and/or non-transparent data transport. QoS can be adjusted for delay, bit error ratio, frame error ratio, etc.
UTRA TDD and TD-SCDMA, also denominated as TD-CDMA, specifications are currently developed within the 3GPP. UTRA TDD has been developed with the UTRA FDD part by harmonizing important parameters of the physical layer and specifying a common set of protocols in the higher layers. TD-SCDMA has significant commonality with the UTRA TDD. Specifications include capabilities for the introduction of TD-SCDMA properties into a join concept. The radio access scheme is a hybrid TD-CDMA. UTRA TDD spreads information over approximately a 5 MHz bandwidth and has a chip rate of 3.84 Mcps. TD-SCDMA spreads information over approximately 1.6MHz bandwidth and has a chiprate of 1.28 Mcps.
The UTRAN consists of a set of Radio Network Subsystems (RNS) connected to the Core Network through the Iu-Interface.
A RNS consists of a Radio Network Controller
(RNC) and one or more Node Bs. A Node B is connected to the RNC through
the Iub interface. A Node B can support FDD mode, TDD mode or dual-mode
operation.
The RNC is responsible for the handover
decisions that require signalling to the UE. The RNC comprises a combining/splitting
function to support macro diversity between different Node B. An RNC supporting
the FDD mode may include a combining/splitting function to support macro
diversity between different Node B. Inside the UTRAN, the RNCs of the Radio
Network Subsystems can be interconnected together through the Iur
Iu(s) and Iur are logical interfaces. Iur
can be conveyed over physical direct connection between RNCs or via any
suitable transport network.
The main difference to the existing GSM
system is the Code Division Multiple Access (CDMA). In principal all cells
and all users share the same radio channel. A basic network with a 5 MHz
radio channel will in theory provide the access. The system is interference
limited, and the services are with a few exceptions limited by the C/I
(signal-to-interference)-ratio.
The main mode (at least in phase 1 of
UMTS) will be FDD (Frequency Division Duplex) with separate frequency-bands
for up- and down-link. In this respect the FDD mode is very similar to
GSM. However, the frequency bands are 1920-1980 & 2110-2170 MHz respectively,
this means radiowave propagation with considerably higher loss. Compared
with traditional GSM-900 approximately 7 dB, and even 1 – 2 dB higher loss
than GSM-1800 must be expected.
In the TDD mode (Time Division Duplex)
the user and the BTS uses the same frequency, but different time slots
for up- and downlink.
As all cells in a UMTS network operate on the same frequency, the system experiences a tight base-to-base coupling. Shadow fading, mulitpath and user respectively service distribution will create a highly variable interference level, which is treated as noise (No).
Some of the key mechanisms used in the CDMA system level simulator are fast power control, soft and softer handover, and load control algorithms. Link-level simulations and system-level simulations will provide a.o. answers to:
A more detailed description of the UTRAN
specifications are given in UMTS
radio architecture specifications ().
Voice service was simulated at 8 kbits/s,
and LCD and UDD services at 64, 144 and 384 kbits/s. The results can be
summarised as follows (for a more detailed analysis see "Results
from link level simulations applied to radio planning of the UTRAN"
().
In case of data service (LCD & UDD), the working point is more sensitive to the mobile speed and to the propagation environment. In particular, in case of 144 kbit/s LCD service, the lower spreading factor (8 against 128 of the voice service) causes problems to the Rake receiver in discriminating those paths with a delay greater than eight times the chip period (» 2 ms). In such a case, a BER floor higher than 10-6 was found with the most time dispersive channels (Outdoor B and Vehicular B) at very low mobile speed (3 km/h). The BER floor has been removed, in the aforesaid cases, using a 6-branch Rake receiver with the last three branches tuned on the paths with delay greater than 2 ms.
Moreover, the absence of the antenna diversity
in downlink causes a sensitive degradation of the link performance, with
respect to the uplink, especially for circuit switched data service.
 |
|
(downlink – Vehicular A channel) |
The link budget is calculated by the following procedure:
1. Uplink path loss evaluation;On the basis of the radio link results the UMTS (FDD component based on W-CDMA access) link budget has been evaluated in case of urban environment. The cell radius of the UMTS system has been compared with the one of GSM 1800.
2. Downlink power level evaluation at cell border;
3. Dowlink EIRP value evaluation per traffic channel;
4. Downlink power evaluation per traffic channel;
5. Downlink path loss evaluation.
 |
|
|
The figure presents results from link-level simulations: The coverage range of UMTS services in urban environment as a function of average power of the mobile station and offered service (70 % cell loading). The UMTS cell radius is compared to the cell radius of the GSM 1800 system. It is worth noting that in the GSM 1800 case the cell radius is not related to the system load.
The results show that, in the case of voice service, the UMTS cell radius is greater than the GSM 1800 one, also in presence of high system load. Concerning the other services, it is not possible to achieve the same coverage as of the GSM 1800 system.
A more detailed discussion of the results
from link level simulations is found in the document "Results
from link level simulations applied to radio planning of the UTRAN
().
In UMTS new technologies are foreseen
to enhance capacity and/or quality in the network. The European ACTS framework
has initiated several projects to develope UMTS and enhance knowledge of
the radio network. Within the ACTS program framework, the STORMS
(Software Tools for the Optimisation of Resources in Mobile Systems)
project developed a network planning tool for UMTS. The planning process
is based on the multimedia type services to be offered by UMTS. The software
tool allows the network designer to calculate cellular network characteristics
like traffic densities, cell sizes and radio coverage. The effects of employing
different access schemes and the location of base station controllers can
be determined. Moreover paging areas can defined. Some of these aspects
were verified through hardware development and tests.
The use of adaptive or smart antennas
allows enhanced capacity and /or quality in cellular radio systems like
UMTS. The technology is described in detail in 
Potential
network architectures for UTRA using SDR-HFR technology. A technique
called the Space Division Multiple Access (SDMA) may be combined with classical
multiple access techniques like FDMA, TDMA and CDMA in order to achieve
a C/I (signal-to-interference) improvement in the range of 15 to 25 dB.
SDMA is an access technique where spatial beam steering is obtained with
an adaptive antenna array, and where Spatial Multiplexing Gain (SMG) is
proportial with the number of antenna elements. A detailed study on the
benefits of the SDMA technique can be found by clicking
here.
Further research areas are multiuser detection, powercontrol based on quality requirements of the applications, enhanced link adaptation, and turbo coding.
The
UTRAN has many features which present new challenges to mobile network
designers. In order to be able to design a successful UMTS Radio Access
Network, it is necessary to have a good understanding of these issues.
While the following is by no means an exhaustive list of the design considerations,
it represents some of the fundamental issues facing network designers.
With many of these issues it is not clear which of the possible solutions
would result in the optimum network design, as there are many dependencies.
The most effective way of evaluating the effects in order to optimise network
performance is through simulation.
The
UMTS Radio Access Network is built around two new nodes and three new interfaces.
The Node B is effectively a UMTS "basestation", while a Radio Network Controller
(RNC) is comparable with a GSM BSC. Each RNC is connected to the Core Network
(both packet and circuit domains) by the Iu interface; RNCs are connected
together with the Iur interface. Each Node B is connected to an RNC by
the Iub interface.
There are some fundamental limits on the
numbers of cells and RNCs that can be supported, due to the way that cells
and RNCs are identified (normally the number of bits in the identities,
but sometimes hidden elsewhere in the protocol definitions). There is currently
no restriction on the numbers of Nodes B in an RNS or PLMN. The hard limits
are as follows:
|
26,435,456
4,096 65,536 No limit defined in the standards No limit currently defined in the standards |
It may be possible to extend the mast/headframe vertically, although this is unlikely, as most will be as tall as possible in order to maximise coverage for GSM. This will have a significant impact on wind loading. More likely would be the availability of space beneath existing antennas. However this will lead to reduced coverage from the UMTS cells, although this will generally be acceptable on capacity cells in urban and suburban areas, where the cell spacing is determined by the traffic load rather than the propagation conditions. Where there is no possibility to install additional antennas (due to space or load restrictions), it may be possible to use multi-band antennas in order to minimise the physical impact of sharing. However, it is likely that such antennas will have reduced performance compared to single band antennas, and so the impact on the link-budget must be assessed to ensure that there is no adverse affect on the coverage of the two systems. Dual band antennas for GSM800 and GSM 1800 are currently available, and several antenna suppliers have plans for dual and tri-band antennas for GSM900/UMTS, GSM1800/UMTS, and GSM900/GSM1800/UMTS. Where there is marginal impact the use of multi-band antennas would be the preferred solution, in terms of cost and environmental impact.
There are two options for the choice of carrier for microcells:
This
may be attractive if an operator has less than three carriers. For certain
deployment scenarios, microcells may be deployed using the same carrier
as that allocated to the contiguous macro cell layer. This technique
has been discussed for use in IS95 CDMA systems [1],
where it is claimed that good capacity increases are possible if the path
loss slope is steeper in the micro cell than in the macro cell. However
the suitability of the technique is seen to depend on the local channel
conditions. For this technique to work most effectively, the macro layer
must be planned with consideration to future micro cell deployement from
the outset. Thus it is unlikely to be suitable for providing hot spot coverage
in an existing macro cell layout, depending on the relative positions of
the other system elements to the hot spots. Also this technique does not
allow traffic to be forced on to the micro layer for congestion relief,
as any terminal operating to the non-prefered cell will cause interference
problems. The resulting system is also very sensitive to handover
performance, requiring a very fast and well set up handover algorithm to
work effectively. Sub-optimal deployment of same carrier microcells
will reduce the overall system capacity, rather than improving it.
The
alternative approach is to allocate separate carriers to the macro and
micro layers in any given area. This allows microcells to be planned in
much the same way as they are currently in GSM. The ability to force terminals
on to the micro layer, despite radio conditions, can then be exploited
to maximise the effectiveness of the micro cells. However this approach
is not without drawbacks. Firstly it requires hard handover between carriers
and the use of compressed mode, with it's associated capacity reduction
due to the increased noise power. Also, where carriers are limited, it
will require a large deployment of microcells in order to see any overall
capacity advantage, unless the traffic in a cell is very localised, since
as the traffic load increases, spectrum would have to be transferred from
the macro to the microcells.
The use of six sectors can lead to an increase in the coverage area that is served by multiple cells (i.e. the soft handover region), depending on the local propagation conditions and the antenna pattern. The two figures show the overlap between the antenna patterns - while this does not match the soft handover regions, it shows how the overlap can increase, given certain antenna beamwidths.
 |
 |
|
| 3-sectors, 90° beamwidth | 6-sectors, 60° beamwidth |
In a practical deployment the amount of overlap would be greater due to the effect of the adjacent sites. Also, the use of six sectors will require much more careful radio planning in order to minimise coverage holes between cells, while minimising the overlap of adjacent cells. In the CDMA system, this overlap causes interference, since every cell is on the same carrier, which in turn leads to a reduction in capacity. The impact of this interference is minimised by the soft handover mechanism, which exploits the overlapping area between sectors as a form of macro diversity. Despite soft handover there is still a net reduction in downlink capacity. Thus it is unlikely that you would experience a doubling in capacity by moving from three to six sectors.
Also, for soft handover to be effective, the receivers must have sufficient Rake fingers to resolve the signals from each serving cell, as well as any multipath signals that may be present. The greater the number of overlapping sectors, the more Rake fingers will be required, and the more complex and expensive the terminals will be. The increased likelihood of soft handover will also have an impact on the load on RNC, which will have to be dimensioned to cope with the increased processing.
Aside from the performance issues of six
sectors versus three sectors, there are the deployment issues to be considered.
The opportunities to deploy six additional antennas on existing sites are
likely to be very limited, due to the limited space and increased wind
loading. The possibilities of using multi-band antennas is also limited,
as the narrow beam antennas would not provide sufficient coverage for GSM,
and using broad beam antennas would lead to even worse interference and
reduced capacity
The link
budget is a fundamental tool for radio network designers. Due to the
nature of the CDMA system, the link budget has a number of different parameters
compared to that for GSM. A typical link budget is provided here, by way
of example. The cells in blue are for user-defineable values.
P920 is entitled “UMTS Network Aspects”. Based on the demand for new multimedia services, the project aims (this project ends in December 2000) at determining architectures and services for global mobility with underlying heterogeneous local access networks. It is relevant to identify candidate technologies. The main concept in the project is the Virtual Home Environment (VHE), defined as the capability to offer a user-profiled service set over a variety of different networks in a seamless and transparent manner. Interesting topics connected to this are the capability of IP to provide secure service delivery and interworking across heterogeneous networks, as well as the development of a generic application protocol providing, among others, the mobility required for UMTS in a VHE.
| Go back to beginning | Do you want to have a condensed version of this information at your fingertips ? Here is a short PowerPoint presentation (in pdf format) to download. |
