Experiments with Local Windows in DAB
Christoph Albinus*, Michael Strey *, Hans-Jörg Nowottne **
*
**
VAD Video-Audio-Design GmbH Dresden
Fraunhofer-Institut für Integrierte Schaltungen, Außenstelle EAS Dresden
Aimed at Local Radio in the SFN a field trial has been carried out in Dresden, Germany. Its
focus was on Local Windows, i.e. the radiation of different data in selected parts of the DAB
transmission frame by different transmitters of the same SFN. The paper reports that localisation can be extended from MSC to FIC as well, leading to advantages like comparable
user interfaces for local services as for global ones and customising of the multiplex according to individual needs of providers.
Laboratory measurements with two fading channel simulators were performed to test the
required co-channel interference protection ratios when using an audio channel or FIC with
different contents.
A new approach, compatible to ETS 300 401, for signalling local services has been created
and tested successfully. It is based on the separation of data, which is exclusively dedicated
to local services, and its frame oriented, cyclic assignment to the FIC. Generation and
analysis of this specific FIC data were done by means of software tools which can be linked
to the real-time equipment by files .
Survey measurements with up to three transmitters and two repeaters were carried out at Lband. Investigations about error rates and audio quality in global and local services depending on field strength and channel impulse response have been made. The coverage area
of local services was determined by mobile measurements.
Conclusions are drawn concerning receiver behaviour and several further aspects.
1.
The DAB field trial in Dresden
Together with VAD Video-Audio-Design
GmbH and FhG IIS/EAS the Deutsche
Telekom AG built up a DAB experimental
single frequency network (SFN) in Dresden.
The SFN consists of three transmitters and
two repeaters radiating at L-Band frequency.
Survey measurements have been carried out
to test so-called Local Windows, a method to
provide different programmes or services at
different (local) service areas of the same
SFN. Normally, this takes place in addition
to the radiation of other (global) programmes
which are identical in the whole SFN. That
means, co-channel transmitters radiate partially different DAB multiplexes. Figure 1
presents the three transmitter sites placed
around the city of Dresden.
759 W ERP
= 4
4,5
µs
Radebeul
13,
4 k
m
20
,3
km
= 6
7,6
µs
209 W ERP
251 W ERP
22,9 km = 76,4 µs
Gompitz
(local window)
Wachwitz
Ri
ve
City of
Dresden
rE
lbe
Repeater
Tharandt
Repeater
Rabenau
Figure 1
Transmitter sites for the DAB field trial
Dresden
1
The field measurements and a series of laboratory investigations were performed by the VAD VideoAudio-Design GmbH supported by FhG IIS/EAS.
2.
Why Local Windows in DAB?
Mainly there are two reasons to deal with Local Windows:
1) a lack of frequencies for DAB ensembles and
2) the need for flexibility using a DAB ensemble.
According to the official frequency allotment plan /CEPT95/ 21 DAB blocks are used for Germany. But
due to the rigid block scheme only two terrestrial DAB blocks are available at each location in Germany
(one at VHF and one at L-Band). On the other hand the coverage areas of the assigned frequencies are not
small enough to eliminate in any case the request for further subdivision into different local service areas.
So it makes sense to have an option for partially reuse of main service channel (MSC) capacity in a SFN.
Concerning the flexibility two aspects should be considered which can be handled by Local Windows.
Firstly, supposing that a service provider simply wishes to split his program into different regional programs dedicated to different areas of the SFN for a limited time. Depending on the sub-channel placement
of these regional services, in the same or in other ensemble, the receiver has to change the selected service
only or the ensemble too. Further supposing the optimistic case that enough MSC capacity is available,
the change can be completely signalled by fast information channel (FIC) data and then can be done
automatically by the receiver based on information about its position. Compared with the effort to implement (and correctly update) the necessary assistance for this service switching, it should be much easier
to substitute the program contents in place regarding to the specific service area. Secondly, by means of
Local Windows it seems to be possible that the ensemble provider assigns a consecutive part of the MSC
to local service providers which in turn define their individual sub-channel and service organisation.
Certainly, Local Windows don’t bring only advantages. One main problem is the corrupted reception of
some services (which can be accepted for local services only) in the interference area due to non-constructive signal superposition between transmitters radiating different DAB multiplexes. These interference
areas are comparable to the reuse distances in the planning of local coverage according to the conventional scheme. Nevertheless the main advantage of Local Windows will be the greater flexibility in planning.
Figure 2 represents a possible situation e.g. in case of DAB along traffic routes.
local service area 1
margin of
coverage for
global services
local service area 2
20
10
0
-10
-20
-80
Figure 2
-60
-40
-20
0
20
40
60 km
region of
interference of
local services
Computation of the coverage with local and global services combined in
one DAB ensemble with eight transmitters.
2
In order to decide whether the Local Window method is applicable or not, first of all the following
questions have to be answered:
1) Is it possible to change the contents of one or more services in the DAB multiplex of a SFN at one
site without causing interferences to the other services of the same ensemble?
2) What about the necessary signalling?
3) What will be the required co-channel interference (CCI) protection ratio if an audio channel with
different programme contents from different transmitters is used? What means this CCI protection
ratio for the extent of the interference areas surrounding one or more local transmitter sites?
3.
Basic experiments: change of OFDM symbols
3.1.
Principle
Figure 3 shows an example of a DAB transmission frame with a localised part in the MSC.
Substitution of OFDM symbols is achieved indirectly by replacing single sub-channels or FIC data in the
ensemble transport interface (ETI) data stream.
When localising a part of the MSC the boundary between a global and a local sub-channel should coincide with the beginning of a new OFDM symbol. In this way a mixture of data bits of local and global
sub-channels during the following OFDM symbol mapping is avoided.
Every change in the contents and structure of the MSC requires an appropriate change in the FIC for the
correct signalling of the services included. In the FIC due to the convolutional coding and symbol mapping the contents of the bit stream of the three successive fast information blocks (FIB) is coupled in the
three OFDM symbols (transmission mode 2) carrying the FIC in one transmission frame. Therefore only
the whole FIC information of one transmission frame should be made either local or global.
Transmission Frame
(Example, mode 2)
24 ms
0,636 ms
57600 bit = 7200 byte = 75 OFDM symbols
Synchronization
Channel
Symbol Mapping,
Partitioning
3 OFDM-symbols
72 OFDM symbols
Fast Information
Channel (FIC)
Main Service Channel (MSC)
2304 bit
Fast Information
Channel (FIC)
55 296 bit
SubCh 1
(140 CUs)
SubCh 2
(140 CUs)
SubCh 3
(140 CUs)
global services
Puncturing
Convolutional Coding
Energy dispersal
SubCh 5
(168 CUs)
SubCh 6
(168 CUs)
local services
beginning of new
OFDM symbol
unused MSC
capacity
768 bit
FIB FIB FIB
Figure 3
SubCh 4
(96 CUs)
DAB transmission frame with localised MSC
3
3.2.
Experimental environment
To confirm the opportunity of Local Windows found theoretically a series of laboratory measurements
with the 4th generation DAB-equipment and two fading channel simulators was performed. Figure 4
gives the block diagram of the equipment for these measurements. The configuration with two PC-recorders having the same ETI data stream from the transport multiplexer as input gives flexibility to replace
any part of the multiplex by data from the hard disk of the two synchronised PC-recorders. Depending on
the actual experiment these were one or two sub-channels of the MSC and in some cases the complete FIC
too.
To deal with FIC data the Generation and Analysis of FIC (GAF) software system developed by FhG IIS/
EAS was used. This tool-set consists of FIC input supplier to define a DAB scenario, FIC assembler, and
FIC analyser. It runs on UNIX workstations and supports the DAB standard /DAB95/ without restriction.
Field strength levels were measured on a Rhode and Schwarz ESVP test receiver. A measure of the
objective reception quality (bit error rate and CRC errors) was obtained from the 4th generation DAB
receiver DAB452 (Philips). All measurements were carried out in DAB transmission mode 2.
276 MHz
252 MHz
TP 1
Oszilloskop
local 1
FIC 1
PCRecorder
(FhG IIS)
CD-Player
COFDMModulator
(BBC)
DAB Receiver
DAB 452
(Philips)
Fading Channel
Simulator
(FADICS)
Audio
Amplifier
CODEC
RDI
DAT-Recorder
T-MUX
(FhG IIS)
1117 MHz
217 MHz
Controler
(R&S PSA)
for BER
Recording
CODEC
PCRecorder
(FhG IIS)
COFDMModulator
(BBC)
Fading Channel
Simulator
(FS 900)
local 2
FIC 2
TP 2
TP 1
or
TP 2
Controler
(MEDAV)
for FIC
Recording
Test Receiver
(R&S ESVP)
750 MHz
DAB Signal Generation
Figure 4
Fading Channel Simulation
DAB Test Receiver
Block diagram of the equipment for the laboratory
measurements
The radio channels between the transmitters with partially different multiplexes and the receiver were
modelled with fading channel simulators. The settings of the two fading channel simulators were obtained
from field measurements with a RUSK X channel sounder /RU92/ in the test area. The so generated
fading corresponds to the model ‘Typical Urban’ used for mobile radio simulations. The driving speed
was set to 50 km/h at a centre frequency of 1465 MHz.
3.3.
Change of MSC symbols
Figure 5 shows the result of a measurement with one local sub-channel exchanged at the end of the MSC.
The CCI ratio was changed step by step by setting the attenuators shown in Figure 4. The x-axis in Figure
5 shows the CCI ratio generated, the y-axis reports the bit error rate (BER) measured at the DAB-receiver
in the sub-channel set. The upper curve in Figure 5a was measured in the local sub-channel (sub-channel
4
6). The lower curve results from measurements in one of the global sub-channels. For comparison Figure
5b shows a curve measured in a sub-channel of a regular DAB SFN multiplex which was identical at both
paths. It’s to be seen that as long as the symbol boundary condition holds there is no degradation in BER
of the global sub-channels compared with the regular SFN. The small increase of BER at low CCI ratios
is caused by degradation of some carriers of the COFDM multi-carrier signal due to multipath environment. This is normal also for the SFN and with usually applied protection levels the DAB system will
tolerate these BERs. With respect to the local sub-channels represented by the upper curve which differs
in the two paths we have to state interferences up to a CCI ratio of 20 dB. At 20 dB CCI we have the same
BER as for 0 dB CCI in the SFN, that means the DAB system will tolerate this BER at 20 dB CCI
without any degradation of audio quality. In the region below this margin disturbances and receiver mutes
are possible. The same experiment was done for ensembles with two or three local sub-channels at different positions in the MSC exhibiting practically the same result.
a)
Ensemble with localised sub-channels
b)
BER 10e-2/s
20
BER 10e-2/s
20
10
10
7
7
4
4
2
2
0
0
30
25
20
15
10
5
0
5
10
15
20
25
30
30
CCI /dB
Figure 5
3.4.
regular SFN
25
20
15
10
5
0
5
10
15
20
25
30
CCI /dB
BER with and without localised sub-channels in the MSC
Change of FIC symbols
For the experiments with localised FIC the FIC was off-line generated using the GAF system mentioned
above. The files (about 6 minutes real-time) were cyclic replayed by the PC-recorder. On this way it was
possible to assemble the FIC data into different frames depending on its local or global assignment. The
FIC information of every fourth transmission became local and described the local part of the MSC, the
rest of the FIC remained global and contained the description of the global sub-channels and non service
specific FIC data as well.
Using a radio data interface board /RDI95/ provided by FhG IIS Erlangen the FIC received by the DAB
receiver could be recorded. After format conversion of the recorded file the FIC analyser tool of the GAF
system could be used to analyse it. So we were able to detect which parts of the FIC were received
correctly and which were not. It was confirmed that in the interference area only the localised part of the
FIC was faulty.
Figure 6 shows the BER of the FIC and the number of CRC errors per 50 frames over both, local and
global parts of the FIC in one diagram. Because of every FIB (see Figure 3) is validated by a CRC and
one transmission frame carries three FIBs, the maximum number of CRC errors possible in 50 frames
will be 150. If the localisation works correctly CRC errors will be allowed only in the localised part of the
5
FIC BER 10e-2 /S
2
0
30
20
10
30 25 20 15 10 5
0
5 10 15 20 25 30
CCI /dB
Figure 6
3.5.
3
1
FIC CRC errors
FIC, that means in case of our experiments (see chapter 4) the three FIBs of
every fourth frame are allowed to be
disturbed. With other words the maximum number of CRC errors in the diagram should be 37.
The result shown in Figure 6 is better
then expected. Due to the high error protection of the FIC the number of CRC
errors not exceeds the number of 30 and
becomes nearly zero at a CCI ratio of
10 dB. That means while driving with
a DAB receiver from the interference
area into a local service covered area
signalling of the local parts of a DAB
multiplex will work earlier than receiving the service itself in a good quality.
BER and CRC errors in the FIC with local parts
Field tests and receiver
behaviour
Field tests in the area of Dresden have been performed with both a DAB 452 receiver and a RUSK X
Channel Sounder. On the test routes the DAB reception quality (Bit errors, CRC errors) and also complex
channel impulse responses for each of the three transmitters were recorded. Field tests were done in
various stages. In a first stage the main part of the RUSK X measurements was carried out. In a second
stage a regular SFN with one multiplex was tested and at least one transmitter in the SFN transmitted a
localised multiplex with both local parts in MSC and FIC.
Results from the extensive field tests of DAB reception confirmed the encouraging test results of the
reception of the remaining global services in the DAB multiplex. No limitations of the reception of global
services in the test area were observed. Coverage of the test area was quite good, but not in all places. It
has to be stressed, that the transmitter network was not yet fully optimized.
Because test area and transmitter sites had not been chosen with respect to a small region of interference
between the different transmitters, places with a CCI of less than 20 dB predominated in the test area.
Reception quality for local services was good in those areas where due to terrain obstructions the signal
from the local site was much stronger or much weaker then the signals from the other transmitters.
CCI needed for a good reception quality depends also on the fading characteristics of the channel. For
fixed receivers (portables) the CCI needed is less than 20 dB. That means, when mobile DAB reception
quality in a car is poor because of a low CCI, with portable DAB receiver one might still have a reasonable good reception quality.
Another point of interest was the behaviour of currently available receivers with respect to Local Windows. The DAB multiplex with differing local parts in the MSC and localised FIC is fully compatible to
the DAB system description /DAB95/. Therefore every DAB receiver should be able to decode the so
changed multiplex in the locally covered regions where the CCI ratio is greater then 20 dB. Indeed the
DAB452 receiver was able to deal with the modified FIC correctly. Moreover, the laboratory measurements have shown that there is no problem in receiving both the global part of the MSC and the global
part of the FIC also in the region of interference. Much more critical was the question, how the receiver
would deal with the partly disturbed FIC information in the interference region. Ideally a receiver should
ignore the disturbed FIBs with the local information and use the correct received global FIBs for decoding
6
the global services. In practice the algorithm used in the DAB452 was as follows: eight subsequent FIBs
must be received without CRC errors, before the MCI is accepted. After the MCI is stored, the receiver
stops analysing the MCI until it is retuned. Therefore in the laboratory it was not possible to retune the
receiver under interference conditions with many FIC CRC errors. During the field tests this problem was
rapidly toned down due to the fading of the mobile radio channel and the high error protection of the FIC.
Retuning failed only one time with the receiver in the still standing car in the middle of the interference
area. After moving for some centimetres the receiver retuned correctly.
4.
An approach for Local Windows in DAB
The basic idea for partitioning of FIC and MIC into global and local parts is illustrated in Figure 7.
Based on the implementation guidelines /IGL95/ the stream of transmission frames can be described as
sequence of 4-frame-blocks, each beginning with a frame carrying the ensemble identification in its first
FIB. Therefore this frame is assigned to carry global FIC data per definition. The FIC capacity of the
other up to three frames can be designed to carry local FIC data. The ensemble provider decides how
many frames are used for local FIC data with respect to estimated signalling needs, i.e. mainly the maximum number of sub-channels contained in the local part. Using the GAF system simulations were done in
order to investigate consequences of the specific FIC partitioning rule. No serious limitations were found.
In case of Figure 7 a ratio of three frames for global to one frame for local FIC data has been chosen. This
was used in the experiments described in chapter 3 too. The actual local-global FIC ratio in terms of
transmission frames will be constant at least up to a possible dynamically reconfiguration of the multiplex. According to /IGL95/ the Multiplex Configuration Information (MCI) has to be assigned to the first
FIB. The FIC data assembling process for local FIC data corresponds to this rule and is therefore the
same as for global FIC data. Consequently, the additional frame assignment rule can be seen as compatible with /DAB95/ and /IGL95/ too.
MSC partitioning in turn is independent of the transmission frame instance. The MSC starts with the
continuous global part and ends at a symbol boundary specified by the ensemble provider. The remainder
of the MSC is called local part and is able to transport a different number of sub-channels depending on
the individual local area.
Compared with the DAB receiver behaviour described in chapter 3.5., we propose to implement a slightly
different behaviour. This is aimed at support so-called implicit reconfiguration which is necessary for the
(mobile) receiver if local service areas are changed. This can be done by the continuous monitoring of the
CRC error numbers registered for the FIC over the last n transmission frames. With respect to the 4frame-sequence decision can be derived whether the receiver is situated inside a local service area or not,
i.e. it is located in an interference area. In the latter case the FIC CRC error number will be significant
increased in transmission frames assigned to local services. Consequently, it is possible to require that the
Transmission Frames
...
G
G
i
G
G
L
G
G
i+1
G
G
G
FIC
FIB1
L
i+2
G
G
G
G
G
G
L
i+3
G
G
L
L
L
MSC
FIB2
Frame 1 contains
FIG 0/0
Figure 7
G
FIB3
global part
local part
G - global FIC data
L - FIC data for local services only
OFDM symbol boundary
Example of a partitioning scheme of FIC and MSC
7
...
receiver mutes if there is no sufficient reception quality due to interferences and vice versa automatically
reconfigures its MSC decoding based on MCI actually received when having reached a new service area.
Be aware that no knowledge about actual receiver location and no signalling of boundaries of local
service areas is needed. This results in reduced effort for ensemble provider and mostly important correct
reaction of the receiver. With respect to the local offset between good FIC reception compared with good
reception of the corresponding service (refer to chapter 3.4.) intelligent trigger levels should be defined.
Further on, it is one of the main tasks for planning of Local Windows in a SFN to customise the interference area. This can be achieved by taking advantage of natural terrain obstructions and by means of
additional transmitter sites (also gap fillers) in combination with directional antennas.
5.
Conclusion and Outlook
We have presented experimental results and, based on them, a new approach for application of Local
Windows to support Local Radio within a DAB-SFN. The approach comprises both an assembling principle for local specific FIC generation and recommendations related to behaviour of DAB receivers. Most
important, it is fully compatible with the DAB standard. Implementation guidelines /IGL95/ can be detailed without any fundamental change.
Our future work is directed to further test the approach under regular operation conditions. Surely that
means real-time insertion of local services and specific FIC generation instead of using prerecorded files.
Therefore the PC-recorder used during the field trial should be developed into a Local Inserter. Corresponding software update of the transport multiplexer is necessary to support specific FIC generation.
Further on, DAB receiver adaptations and tailoring of coverage areas have to be tested. Moreover, connection of distributed providers by public networks and management has to be taken into account, e.g. use
of Service Transport Interface (STI).
Finally, integration in a DAB pilot project, e.g. in Saxony next year, is under discussion.
Acknowledgements
This work was supported by Sächsische Landesanstalt für privaten Rundfunk und neue Medien.
Thanks also to all of the colleagues from FTZ der Telekom Darmstadt and Berlin, PlR Baden-Baden and
FhG IIS which have contributed to this work.
References
/CEPT95/
/DAB95/
/IGL95/
/ETI95/
/RDI95/
/RU92/
/RU93/
CEPT/ERC, T-DAB Planning Meeting, Wiesbaden, July 1995
ETSI, Standard ETS 300 401, Digital Audio Broadcasting (DAB), February 1995
DAB-System: Guidelines for Implementation, EUREKA Project 147, March 1995
Definition of the Ensemble Transport Interface (ETI), Version 4.0, EUREKA Project 147,
March 1995
DAB-System: Preliminary Specification of the Receiver Data Interface (RDI), Issue 1.3A,
EUREKA Project 147, WG D- RDI Task Force, November 1995
U. Martin, H. W. Schüßler, K. Schwarz: A Device for Propagation Measurement in Mobile
Radio and a Post-Measurement Modelling Procedure. Frequenz, Vol. 46, 7/8, 1992.
R. W. Lorenz: Outdoor Wideband Mobile-Radio Propagation Studies in Europe. IEICE
Trans. Commun., No. 2 February 1993.
8