ON-AIR MULTIPLEXED UPLINKING OF EUREKA-147 DAB TO EMS

ON-AIR MULTIPLEXED UPLINKING OF EUREKA-147 DAB TO EMS R.H.Evans & S.T.Baily BBC, UK Abstract Digital audio broadcasting via satellite, using the Eureka-147 system, is seen by many as the future replacement for short-wave radio, providing high-quality audio and data services to mobile receivers worldwide. However, international broadcasters' studios are at widely separated locations, and this creates a problem of how the individual contributions from many broadcasters are combined into a single multiplex for transmission to the consumer. This paper describes how on-air multiplexing using synchronised earth stations can overcome this difficulty, allowing a standard Eureka-147 signal to be broadcast from a conventional satellite, without resorting to on-board processing techniques. On-air multiplexing is applicable to both GEO and HEO satellites. Introduction International broadcasters have traditionally used short-wave transmissions to deliver radio programmes to listeners around the world, and while this provides a large coverage zone the audio quality can be poor. Digital broadcasting via satellite can provide high quality audio to both mobile and stationary receivers, over large areas, and can also carry a variety of other data services. The Eureka-147 system [1], which uses the COFDM [2] transmission technique, is becoming the world-wide standard for terrestrial broadcasting. In recent years, it has been shown that the system is also suitable for satellite delivery, (via tests on the Optus B3 and Solidaridad 2 satellites [3]). More recently, in March 1997, BBC R&D Department, in conjunction with Nuova Telespazio and ESA, produced the first European demonstration of satellite DAB to delegates at the EBU headquarters in Geneva using the European Mobile Services (EMS) payload. Multiplexing Options Eureka-147 is based on a programme multiplex and can carry a combination of many separate services, each at a variety of bit rates. However, these services will probably originate from broadcasters' studios at widely separated locations, and this creates the problem of how the services can be brought together into a single DAB multiplex or "ensemble" for transmission to the consumer. The simplest option would be to use a single central "hub" earth station fed by long distance telecommunications links. However with reliability issues, ongoing costs, and the political situation outside the broadcaster's control this may not be popular. Broadcasters may, therefore, prefer to uplink their audio channels from an earth station located at their own premises, and under their own direct control One possibility for local uplinking would be to use an On Board Processing (OBP) satellite, and assemble the multiplex in the satellite itself. However, cost, reliability and flexibility issues are a major concern with OBP. A far more attractive option is the possibility of using a

conventional satellite with its "transparent transponder" architecture and a Time Division Multiplexing (TDM) technique, which is the method described in this paper. The TDM approach will allow each local earth station to uplink its audio service to a conventional satellite by directly contributing a few correctly timed COFDM symbols to the DAB frame. By synchronising the transmissions from each earth station it is possible to create a composite signal at the satellite's input antenna which conforms exactly to the Eureka-147 standard. With no requirement for telecommunications links, or a dedicated OBP satellite, and freedom to uplink from any location visible from the satellite, this TDM system would be most attractive. Although applicable to both GEO and HEO satellites, this paper will focus on the GEO application. Fig 1 - TDM Uplinking via local earth stations. DAB and the COFDM system The COFDM transmission system used for Eureka-147 DAB is unlike the majority of digital transmission systems. It uses a very large number of orthogonal carriers (typically hundreds), each modulated independently at a relatively low symbol rate, as shown in Fig. 2. The overall wide bandwidth, together with frequency and time interleaving, and the use of a guard interval provide a very rugged transmission system which is particularly suited to mobile reception. Fig. 2 - The Eureka-147 signal in the frequency domain (Mode III). The Eureka-147 DAB system was designed to operate in one of four modes, according to the frequency band used for transmission. Each mode uses a different number of QPSK carriers, but an overall bandwidth of 1.537 MHz. This paper concentrates on L-band transmission using mode III parameters, which specifies 192 carriers, with each carrier modulated at 16 kbits/sec, and an inter-carrier spacing of 8 khz. Mode III is suitable for use at up to 3 GHz and, following WARC92, satellite digital audio broadcasting (generically) has been assigned spectrum in L-band from 1452 MHz to 1492 MHz

The DAB transmission frame Looking at the Eu-147 signal in the time domain, it is made up of transmission frames of 24 ms duration, as shown in Fig. 3. Each frame consists of 154 symbols -- the null symbol for coarse synchronisation of receivers, the phase reference symbol for fine synchronisation and tuning, the 8 Fast Information Channel (FIC) symbols which carry (amongst other things) the Multiplex Configuration Information (MCI), followed by 144 symbols of payload data which form the main service channel (MSC). The MSC carries the audio data, coded using the MPEG Layer I standard, and is divided into sub-channels with each sub-channel carrying 24 ms of audio for a particular service. Fig. 3 - The Eureka-147 frame structure (Mode III). The symbol duration is 156 s, but due to the multi-carrier nature of DAB each symbol carries 384 bits. Every symbol allocated to a sub-channel therefore occupies 16 kbits/sec, so for example, an audio channel coded at 128 kbits/second and rate 1/2 FEC would occupy 16 symbols. The position of each sub-channel within the transmission frame is fixed from frame to frame, and this allows the multiplex to be created from separate contributions for the purposes of uplink multiplexing. Fig. 4 shows how several contributing earth stations could create such a composite multiplex in this way.

Fig. 4 - Building a TDM composite signal from individual contributions (e.g. 128 and 64 kbits/sec). The first contribution of the frame will carry the phase reference symbol and the FIC, and presumably (though not necessarily) at least one sub-channel. This is designated as the primary earth station, and sets the frequency, timing and power levels. The secondary earth stations each transmit only their own contribution(s), the format of which must agree exactly with that specified in the MCI which is carried in the FIC and only transmitted by the primary earth station. Eureka-147 uses differential QPSK modulation, and at the hand-over point between contributions there will be a phase discontinuity preventing the first symbol of every secondary contribution from being decoded. The first symbol effectively provides a new phase reference, allowing the second and subsequent symbols of each secondary contribution to be decoded as normal. The first symbol cannot, therefore, be used to carry any user data and this may be a little inefficient, but in mode III it is not a great problem, as each lost symbol amounts to just under 0.7% of the total user capacity. An arrangement using 10 geographically separate uplink sites (i.e. 1 primary and 9 secondary earth stations) would therefore reduce the available data capacity by only 6.25%. In practice an allowance is made for this symbol in the secondary earth station's sub-multiplexer unit. It is simply configured to insert a symbol's worth of random data at the start of each contribution, before the actual audio service component. However, as this additional symbol is not specifically identified in the multiplex configuration data it does not cause any problems to the listener's receiver. Hardware In order to prove the concept of TDM uplinking, two complete transmission chains have been constructed at BBC R&D Department. Each chain comprises of audio coders, a sub-multiplexer, a COFDM generator, a proprietary switching unit, and a GPS receiver, as shown in Fig. 5.

Fig. 5 - TDM Uplinking System block diagram. In each transmission chain the sub-multiplexer is synchronised directly to the 1 Hz reference signal and also the time code from the GPS receiver. The output of the sub-multiplexer is then sent to the COFDM generator where it first passes through a FIFO buffer. This FIFO delay is set to compensate for the differences in the path lengths between an individual earth station and the satellite. In the current implementation of the TDM uplinking system, the COFDM generator produces a continuous DAB signal throughout the duration of the transmission frame, and its RF output is then switched on/off in order to select only the symbols required for the local contribution. The control signal from the switching unit derives its timing from the I/Q output of the COFDM generator and so experiences the same buffer delay as the COFDM signal (rather than being in phase with GPS time). The resulting bursts of contribution data occurring every 24 ms are then passed to the uplink equipment for transmission in the conventional way. If all the earth stations are synchronised, and the appropriate delay is set in the COFDM generator, the uplinked contributions will then arrive at the satellite's input antenna in the correct order and at the exact time required to construct the standard Eureka-147 multiplex, on-air. Creating a composite multiplex Although this composite signal will be the combination of data bursts from several different earth stations, it must not exhibit any artefacts of its TDM origination. The three fundamental parameters which must be kept as constant as possible are relative timing, uplink frequency and power level. Each of the impairments has the potential to degrade the transmitted signal, and in combination the effect will be additive. A series of tests was performed in the laboratory at Kingswood Warren to assess the effect of each The transmission chain equipment. parameter in isolation. These tests used the two complete transmission chains to represent a primary and a secondary earth station. Transmission synchronisation Unlike conventional TDMA systems, in which data bursts transmitted through the satellite are generally unrelated, and intended for different receivers, the TDM uplinking system aims to combine the data bursts into a single composite signal. Whereas TDMA requires a short guard time between data bursts, the time between TDM contributions should ideally be zero. The

data bursts must be accurately synchronised so as not to create any overlaps or gaps in the signal at the hand-over points. The curve for synchronisation error v BER (Fig. 6) shows a negligible degradation for timing errors between approx. 15 s s and +15 s, which is to be expected for a guard interval of 31 s. Outside this region the BER increases steeply, and at approximately +/- 20 s impairments to the decoded audio will be heard. Fig. 6 - Synchronisation v BER of secondary contribution. While the synchronisation accuracy required for a satellite-only delivery system would therefore be a few microseconds, the use of co-channel terrestrial gap filler transmitters requires that the COFDM guard interval should not be reduced. Therefore in a real system we should aim for a figure of around 1 s. Using a specialised GPS timing receiver we can expect to achieve an accuracy of better than 300 ns for most of the time. Slant path length compensation In a similar way to TDMA operation, an allowance needs to be made for the fact that each earth station will be located at an almost arbitrary point on the Earth's surface. The path length between each earth station and the satellite will, therefore, be different, and a compensating delay is required so that the data bursts arrive at the satellite at the appropriate time and create a seamless composite DAB signal. The maximum difference in the slant path delay between any two earth stations is readily calculated, being between 0 and 18 ms. This fixed difference is compensated for by the buffer delay built into the existing COFDM generators, which is adjustable in increments of approximately 1 s. Satellite station keeping errors Several factors affect the slant path distance (including the irregular ellipsoid geometry of the Earth and even the earth station's height above sea level), however, the effect of orbital drift means that an accurate path length calculation is not amenable to a simple fixed formula. For a satellite with a station keeping tolerance of +/- 0.05 degrees orbital drift may cause significant differential delays depending on the location and separation of the contributing earth stations. While the predictive method of satellite tracking could provide a solution of adequate precision, it may be preferable to employ a more self reliant method such as monitoring the timing error of the composite broadcast signal received at each secondary earth station.

Frequency matching of earth stations In a single uplink (non-tdm) application, the up-conversion oscillators need not be particularly stable as the receiver's AFC is capable of compensating for some error. However, in the TDM uplinking system, the receiver's AFC and phase reference circuitry operate only on the phase reference symbol at the very start of the DAB frame, and therefore will only "tune in" to the contribution from the primary earth station. Switching to a different signal (i.e. a secondary contribution) part way through the frame means a possible step change in the frequency, and any frequency difference here will degrade the signal. Up-converter frequency stability is therefore a key issue. Tests in the laboratory using the dual transmission chain show that a negligible degradation to the signal occurs for an error of around +/- 200 Hz, (using the Mode III inter-carrier spacing of 8 khz) as can be seen from Fig. 7. This corresponds to a requirement for Ku band accuracy of 1 part in 10 8 or better, which is not difficult to achieve. Fig. 7 - Frequency matching v BER of secondary contribution. Power level matching Laboratory tests also showed that an imbalance between the output powers of the primary and secondary transmitters was not in itself a problem, and that the system was able to withstand considerable variation in envelope amplitude through the transmission frame. However, for a downlink power limited system such as this, the satellite must still operate at its optimum power output throughout the transmission frame simply to maximise the received signal level. Using the transparent transponder approach means that gain compensation for incorrect uplink power levels will not be possible at the satellite, so each contribution should be matched to within a fraction of a db when it arrives at the satellite's input antenna. The signal levels received at the satellite will depend on several factors -- such as antenna gain variation, local weather and spreading loss. The envelope of the broadcast signal would need to be monitored at each secondary site to ensure equal amplitude between the local contribution and that originating from the primary site. This would take into account all of the variables in the uplink chain. Fault tolerance As with any time division multiplexed system, a lack of data at the appropriate time, or a

major overlap between two contributors, will lead to some disruption of the service. However, it was found that the Eureka-147 TDM system is surprisingly tolerant of fault conditions. During periods of gross disruption to the frame structure receivers are generally able to maintain uninterrupted operation, providing that the primary earth station continues to transmit the phase reference symbol and FIC satisfactorily. Any failure of a secondary earth station only affects those listening to the contributions from that particular earth station. Listeners to other services on the same multiplex will be unaffected. In an operational system occasional outages are inevitable, and in the event of a total failure of an earth station, it would be normal practise for a standby facility to take over until the fault was rectified. The ability of the system to tolerate frame structure disruption means this back-up approach would be completely transparent to most users. Satellite Tests Having carried out a thorough feasibility study, both theoretically and in the laboratory, we are confident that the TDM uplinking system will be work effectively. In order to demonstrate this, the next stage of the work is to test the system over a real satellite path using two earth stations separated by a large distance, and operating with independent synchronisation. We hope to produce up to date results of these tests during the presentation of this conference paper at ECSC-4 in Rome. Developing TDM uplinking One of the advantages of multiplexed uplinking is that the system can evolve without the listener being aware of any changes. Whatever the source of the Eu-147 multiplex, to the receiver it should all appear the same. This would allow a broadcaster to switch from using a hub earth station (fed via a terrestrial telecomms link) to using their own local facility almost seamlessly, and at their convenience. The main advantage with the TDM approach is that it allows local uplinking, without the need for any form of onboard processing. However, if in the future it is decided that OBP is indeed a viable option then this would be a totally compatible upgrade path. While the uplink multiplexing technique is undoubtedly most easily applied to a satellite in geostationary orbit, the concept could readily be extended to a satellite in a Highly Elliptic Orbit (HEO), e.g. the Mediastar system proposed by DASA. This would require more sophisticated compensation for time and doppler variation, both of which would be constantly changing with orbital position. Conclusions There is considerable interest in applying the Eureka-147 Digital Audio Broadcasting system to satellite delivery of international radio services. However, one of the operational requirements for such a DAB system is for a broadcaster to be able to uplink from their own local earth station facility. Local uplinking of a contribution to a multiplexed service would normally require the use of an onboard processing (OBP) satellite with its additional expense and inflexibility. Time Division Multiplexed (TDM) uplinking as described in this paper would allow uplinking from a local earth station without the requirement for an OBP satellite. Instead, a 'general purpose' transparent transponder satellite would be used. One of the key issues in the TDM uplinking scheme is the synchronisation of each of the earth stations, and this can be readily achieved using a GPS reference, at relatively low cost. Each uplink site must be provided with a compensating delay so that all of the contributions arrive at the satellite at the correct time, irrespective of the uplink sites' geographical locations and slant path lengths. The facility to provide this delay is already implemented in all Eureka-147

DAB COFDM transmitters. While the TDM uplinking technique is most easily applied to a geostationary satellite, it would also be applicable to HEO constellations such as the DASA Mediastar proposal which is aimed at providing S-DAB services in Europe. References 1. ETSI, 1996. Radio broadcasting systems; Digital Audio Broadcasting (DAB) to mobile, portable and fixed receivers, ETS 300 401. 2. Shelswell, P., 1995. The COFDM modulation system: The heart of digital audio broadcasting. Electronics & Communications Journal, 7(3), June. 3. Zubrzycki, J. et al, 1996. Experimental satellite broadcast of Eureka 147 DAB from Solidaridad 2. BBC R&D Report 1996/5.