Method for broadcasting a data stream in a network including a plurality of transmitters, computer software product, head end and system for implementing said method

Translation of description of equivalent WO 2009103638 A1

PROCESS FOR RELEASE OF DATA FLOW IN A NETWORK INCLUDING A NUMBER OF ISSUER AND PRODUCT COMPUTER PROGRAM, NETWORK HEAD AND SYSTEM FOR IMPLEMENTING THE METHOD

1. Field of the Invention

5 The field of the invention is that the transmission and dissemination of digital information, including radio or television data in a communication network comprising a plurality of transmitters.

More specifically, the invention relates to the synchronization of a network 10 such issuers.

The invention applies more particularly, but not exclusively, to SFN networks ( "Single Frequency Network" to "Single Frequency Network"), regardless of the broadcast standard used:

- DVB-T or DVB-T2 (in "Digital Video Broadcasting - Terrestrial 15" in French "Digital Video Broadcasting - Terrestrial");

- DVB-H (in "Digital Video Broadcasting - Handheld", French "Digital Video Broadcasting - Portable");

- DAB (in "Digital Audio Broadcasting", in French "DRB");

20 - DMB (in "Digital Multimedia Broadcasting" in French

"Digital Multimedia Broadcasting");

- WIMAX (in English "Worldwide Interoperability for Microwave Access";

- Etc.

25 For example, the invention is in the context of the ISO / IEC

13818-1 on multiplex MPEG-TS ( "Motion Picture Expert Group-Transport Stream").

2. Background Art

It focuses more particularly in the remainder of this document to describe a 30 existing problems in the field of broadcast networks Digital Terrestrial Television. The invention is of course not limited to this particular field of application, but is of interest to any technical transmission or dissemination of information to deal with a relative or similar problem, particularly in broadcast networks putting implement the standard DVB-T or DVB-H.

digital terrestrial broadcast networks, also known as TNT, implementing DVB-T or DVB-H, are now deployed in France, in Europe and in several other states of the world. In the vast majority, these networks are of the MFN ( "Multi Frequency Network" to "multi-frequency network"), which means that different issuers of such a network operate at different frequencies. Conversely, in certain geographic areas, networks are of the SFN, or isochronous, meaning that individual issuers must be precisely synchronized in time, frequency and content. Indeed, the principle of operation of such SFN may include transmitting one signal from at least two geographically separate sites on each of which is implanted a transmitter. The aim is then to add the contribution of these two reception signals, which requires that they be received at the same time, in a guard interval which is a function of modulation profile and proportional to the width of the symbol time and at the same frequency, to prevent them from disturbing the other. Depending on the geographical distance from the receiver to each of the two transmitters, it is sometimes necessary to take into account different journey times of the signals, and more generally of the propagation channel and disturbances that may introduce.

Because of this need for time synchronization and frequency of the different transmitters, the implementation of such SFN networks proves particularly tricky.

Several methods have, to date, been proposed to enable sync DVB-T signals output by the transmitters of a broadcast network TNT, and are mostly based on a time stamp of the data frames to be transmitted, as presented below in connection with Figure 1. Such a labeling method is standardized, and one can refer for further information this process the referenced standard ETSI TS 101 191. Figure 1 shows, according to this standard, a block diagram of an SFN-type digital terrestrial broadcasting system implementing a data release in MPEG2-TS format ( "Motion Picture Expert Group - Transport Stream").

In this figure 1, there is shown two emitters 10 and 11, each comprising a synchronization device (SYNC) 101, 111, and a DVB-T modulator 102, 112. The synchronization equipment 101, 1 1 1 is fed by both frequency and time reference signals, for example a signal corresponding to one pulse per second, or 1 pps (for "pulse per second"), and a 10-MHz signal resulting from the 1 pps. It is noted that there are exactly 10 million periods of the frequency reference signal at 10 MHz between pulses 1 pps.

These signals can be from any reference system 105 and 115 robust, eg the American GPS positioning system ( "Global Positioning System") or European Galileo or radio carriers large wave (DCF77 in Germany, MSF UK France Inter in France, etc.) - trademarks.

The frequency reference signal of 10 MHz can also be used at the headend, for an SFN adapter, to calibrate its output rate, so that it is accurate and stable, as well as the emission center in transmitters for adjusting their debit.

This reference can also be used by issuers to synchronize their transmission frequency, which must be accurate to within 1 Hz near TNT in an SFN network for optimum operation.

The data to be broadcast by each of the transmitters 10, 11 are received in the form of an MPEG-2 TS transport type (for "MPEG-2 Transport Stream "), from a receiver 12 also acting as a network adapter (" RX network adapter ").

Upstream, at the other end of the broadcasting channel, the MPEG-2 TS data to be transmitted is constructed by an MPEG-2 multiplexer referenced 13, which performs framing of the data. Such an MPEG-2 multiplexer is for example in a national network head, from which are then transported by satellite (in a transmission system, also called a distribution system) the data to be broadcast by each of the transmitters 10, 11 , the broadcast network. After MPEG-2 multiplexer 13, the data is processed by an adapter SFN 14, which sends the timestamp of frames from the same time and frequency reference system 15 as 105, 115 that is used by the synchronization equipment 101, 1 1 1 10 transmitters and 1 1. the SFN adapter 14 is the counterpart to the issue, the synchronization equipment 101, 111 for reception. Thus, the SFN adapter is also supplied with a frequency reference signal at 10 MHz and a temporal reference signal at one pulse per second.

The output of the SFN adapter 14, the data stream is MPEG-2 TS kind: it is then transmitted via a network adapter 16 ( "TX Network Adapter"), and conveyed via the transmission network 17 or distribution (eg a satellite distribution network), to the receivers 12, to be made available to the transmitters 10 and 11.

More specifically, the time stamp carried by the SFN adapter 14 consists on the one hand, to build mégatrames, each corresponding to 8 DVB-T frames 8K mode, or 32 DVB-T frames in 2K mode, and on the other hand, to insert at any position of each of these mégatrames, an initialization packet mégatrame, also called "Mega-frame initialization packet" or MIP.

The MIP packet of index mégatrame n, denoted MIP_not, Is identified by its own PID (for "Packet Identifier" or "packet identifier") and includes: a two-byte word called "point", which gives the number of data packets (TS packets) between the current MIP and the first TS packet in the following mégatrame; a three-byte word called "synchronization time stamp", or STS, which gives the number of periods of 10 MHz between the last pulse

I pps previous reference the beginning of the index mégatrame n + 1 and the beginning of the next mégatrame of index n + 1 (identified by the first bit of the first packet of this mégatrame).

Figure 2 shows more precisely the different concepts for: - the data stream output by the SFN adapter headend, referenced 21;

- The input data stream of the modulator on the transmission site, referenced 22; and

- The modulated signal broadcast by the transmitter, referenced 23. As previously reported, the SFN adapter 14 organizes the data stream 21 mégatrames, and inserts and one MIP packet mégatrame (MIP_not-I for mégatrame n-1, MIP_not for mégatrame n).

At the emitters 10, 11, 101 SYNC module 102 receives at its input, firstly the MPEG stream 22 that was transported in the network, and also the temporal references 1 pps and frequency 10 MHz for example from the GPS receiver.

II research MIP package_not-I.

Having found the MIP package_not-I found it with the value "point" the first TS packet of the next mégatrame n referenced TS_notJ. was therefore carried out the synchronization at the bit level.

Having found the first TS packet_notJ, the system SYNC module 101, 102, located through the STS value and 1 pps pulse, when the first TS packet in the following mégatrame left the SFN adapter headend. This is the time (or time) of transport. Finally, SYNC 101 module 102 deduces the moment distribution, which corresponds to the time of release of the SFN adapter headend with an added delay and mastered common to all issuers of emission sites (maximum or "Max Delay", also carried in the packets PID), and a delay which may be specific to each transmitter ( "Tx_time_offset").

In other words, the emitters 10, 11 use the MIP signaling and a time reference (e.g., a 1 pps signal) identical to that used at the transmitter of the headend, to make a comparative analysis of the signaling MIP and STS time stamps, and make the decision to delay more or less the frame received, the output of the transmitter of the transmitting site. Thus, the deterministic method, which is based on the same time reference 1 pps that headend, provides time synchronization of signals output by the transmitters of the transmission sites.

However, this synchronization is only possible if firstly the transmission time is less than the Max Delay value (itself less than one second if the time reference is used in 1 pps), and the other hand if the time reference 1 pps "draw", that is to say, emit a pulse at the same time in different emission sites.

It is therefore necessary that the temporal references 1 pps and frequency 10 MHz are common in all the points of the distribution chain. We therefore deduce the conventionally a GPS reception.

To reduce equipment costs, it is seeking to develop SFN networks do not rely on the use of GPS receivers for each transmitter site. For example, WO 2006/084361 proposes to include time information in the stream of data to be broadcast at the headend, to retrieve this information at each transmitting site, and generating, from this time information, a reference signal used by the different emission sites to synchronize. However, this technique ensures the synchronization of different issuers only if all transmission sites use the same synchronization technique. Therefore, this technique does not provide proper synchronization of all the transmitters if some transmitters use a reference 1 pps generated from a GPS type receiver, and other issuers regenerate a reference signal from the information temporal scope in the broadcast stream.

Indeed, due to the variation of the position of the satellite used to carry the data stream (MPEG-TS) of the headend to the different emission centers, the transport time between the headend and the various emission sites varies. For example, the transport time to a particular issuer center varies over time of about 270μs, considering a geostationary satellite located about 36,000 Km whose position varies in a cube of about 80 Km from the side.

The time reference 1 pps regenerated according to the technique of WO 2006/084361 is no longer synchronized with the time reference 1 pps after a GPS receiver. In other words, there is a gap of Δlpps between the time reference 1 pps regenerated according to the technique of WO 2006/084361 and the time reference 1 pps after a GPS receiver, at two transmitters of the same plate SFN. In addition, the satellite moves induced frequency variations.

More specifically, the satellite motion and speed induce, due to Doppler, jitter on the flow rate, and thus the frequency reference at 10 MHz. Therefore, the transmission frequencies of the different sites are likely to have a difference of a few hertz. It therefore appears impossible to SFN, within the same network, issuers whose time references are not common in all parts of the distribution chain.

3. Summary of the Invention

The invention proposes a novel solution which does not have all of these drawbacks of the prior art, as a method for broadcasting a data stream in a broadcast network comprising at least two separate transmitters powered by a headend, said stream being organized into data frames and comprising at least one time marker. According to the invention, such a method implements the following steps, at the headend:

- Obtaining a first time reference from an external source;

- Obtaining a second time reference from said data stream received by said headend;

- Comparing said first and second time references, to determine a time offset between said first and second time references;

- Transmission of the time lag or at least one time marker modified according to the time offset, so as to compensate a variation in transport time between said headend and said emitters.

The invention thus provides, according to at least one of these embodiments, a new synchronization technique of the different transmitters of a broadcast network, not requiring the use of a reliable time reference (obtained for example from a GPS receiver) at each of these transmitters.

This technique thus makes it possible to use, in the same broadcasting network, both transmitters using a time reference from a reliable source, and transmitters using a time reference from the data stream. For example, the first time reference is a pulse per second, following a positioning system, and the second time reference is a pulse per second, regenerated from the data stream.

Specifically, here means a reliable source and time reference source or frequency, type American GPS positioning system and the European Galileo radio carriers or long wave (DCF77 in Germany MSF UK, France Inter in France, etc.) - trademarks.

To do this, a time offset is determined between the two time references obtained in different ways, at the headend, and this time shift is transmitted, or at least one time stamp of the data streams changed according to this shift , to the issuers.

Specifically, the time lag determined for example for a mégatrame index n, can be taken into account for the change or time stamps later in the data stream, that is to say, the time stamps of mégatrames index n + 1, n + 2 and the following index. This modification of the temporal markers allows to postpone this time lag, due mainly to the change in the transport time between the headend and transmitter (due to variations satellite), on the modulated signal broadcast by the transmitter.

It can be considered that the transport time between the headend and the various stations of the same city is the same, and that the transport time between the headend and the various issuers within a country is almost identical (difference of less than 5μs).

It is also possible to transmit the time shift to different transmitters, which will use it to compensate for the variation in transport time between the headend and transmitter.

In particular, the time stamp carries a temporal synchronization stamp (STS).

According to one embodiment of the invention, the marker (s) time (s) changes (s) has (have) a value representing the time difference and said timestamp synchronization (STS).

In other words, the time markers are modified by adding the time difference to the value of the timestamp.

According to a second embodiment of the invention, the broadcast system includes a predistorter step of the first time reference, by adding said time offset. Or the marker (s) Time (s) modified (s) then are a value representative of the time reference predistorted and said time stamp synchronization (STS).

In other words, the modified timestamps are the value of said stamp changed according to the time reference predistorted.

Moreover, the diffusion process may comprise a step of obtaining a frequency reference from said first time reference and a predistortion step of said frequency reference, based on said time offset. According to a third embodiment of the invention, the diffusion process includes a step of determining a time reference from said predistorted frequency reference.

Or the marker (s) Time (s) modified (s) then are a value representative of the time reference predistorted and said time stamp synchronization (STS).

The value of said stamp is changed according to said time reference predistorted.

According to a fourth embodiment of the invention, the time offset is transmitted to said emitters by means of an external transmission medium.

For example, this means external transmission implements an external link as a link GPRS, WIMAX ... using a data transmission protocol such as IP.

The time difference or said at least one modified time stamp may also be inserted in the stream of data prior to the transmitting step.

For example, the time lag or said at least one time marker is inserted by replacing a padding packet of said data stream by a phantom packet prior to the transmitting step. Alternatively, the time lag or said at least one marker time is inserted by replacing a padding packet of said data stream a packet including the signaling does not change the nature of the flow, that is to say, the final service associated with the stream.

In particular, this or these specific packet marking inserted into the stream before transmission can be filtered before the extraction of time stamps, that is to say at each transmitter, for example by crushing these packages again by a padding packet.

It may be noted that filter specific packet marking is deterministic. Thus, as part of an SFN network type, it does not change the appearance of the SFN broadcast signal.

In a particular aspect of the invention, the data frames are organized in mégatrames, each mégatrame comprising a timestamp, and the timestamp is an initialization packet mégatrame (PID).

It is well placed in the context of a transmission of a data type of MPEG-TS stream, mainly in the field of type of broadcast networks DVB-T or DVB-H.

Preferably, the network is an SFN-type network, wherein each of said transmitters use a same transmission frequency.

The invention also relates to a computer program product downloadable from a communications network and / or recorded on a computer readable medium and / or executable by a processor comprising program code instructions for the implementation of the diffusion process described above.

In another embodiment, the invention relates to a headend supplying at least two separate transmitters, in a broadcast network to a data stream, said stream being organized into data frames and comprising at least one time marker.

According to the invention, the network head comprises:

- Means for obtaining a first time reference from an external source; - Means for obtaining a second time reference from said data stream received by said headend;

- Means for comparing said first and second time references, to determine a time offset between said first and second time references; and

- Transmitting the time shift means or at least one time marker modified according to the time offset, so as to compensate a variation in transport time between said headend and said emitters. Such headend is particularly adapted to implement the diffusion process previously described.

In particular, such a network head comprises modifying means or the said time markers, depending on the result of said comparison. In this case, the comparison means and of the temporal markers modifying means may be included in a type of adapter

SFN.

Yet another aspect of the invention relates to a system for transmitting a data stream in a network comprising at least one headend supplying at least two separate transmitters, said stream being organized into data frames and comprising at least one marker temporal, wherein: said at least one network head comprises:

- Means for obtaining a first time reference from an external source; - Means for obtaining a second time reference from said data stream received by said headend;

- Means for comparing said first and second time references, to determine a time offset between said first and second time references; - Means for transmitting said time lag or at least a timestamp changed depending on said time difference; at least one of said transmitters comprises:

- Means for receiving said flow data and said time shift, or said data stream comprising at least one modified time stamp;

- Means for determining a time reference from said external source;

- Synchronization means of said data stream from said time reference; - The means to reissue said data stream; and at least one other of said transmitters comprises:

- Means for receiving said flow data and said time shift, or said data stream comprising at least one modified time stamp; - Means for determining said second time reference from said data stream;

- Means for synchronizing said data stream from said second time reference;

- The means to reissue said data stream; whereby said data streams transmitted by each of said transmitters are synchronized.

The proposed technique thus makes it possible to use, in the same broadcasting network, both transmitters using a time reference from a reliable source, and transmitters using a time reference from the data stream.

4. List of Figures

Other features and advantages of the invention will become more apparent from reading the following description of a particular embodiment, given as a simple illustrative and not exhaustive, and the accompanying drawings, wherein: Figure 1, already described in relation to the prior art, shows a block diagram of a digital terrestrial broadcasting system SFN type of implementing a data distribution format MPEG

2; - Figure 2, also described in connection with the prior art, illustrates in the form of timing diagrams, the principle of the SFN synchronization implementation in the system of Figure 1; Figure 3 shows the main steps of the diffusion process according to the invention; - Figures 4 A and 4B illustrate the regeneration of a time reference from the MPEG data stream; 5 shows a technique for regeneration of a frequency reference from the MPEG data stream; Figures 6 and 7 provide a block diagram of a headend and a timing diagram illustrating a first embodiment of the invention; Figures 8 and 9 provides a block diagram of a headend and a timing diagram illustrating a second embodiment of the invention; - Figure 10 shows a block diagram of a head according to a third embodiment of the invention; Figures 11 and 12 show a block diagram of a headend and a transmitter at a transmission site according to a fourth embodiment of the invention; - Figure 13 shows the simplified structure of a network head by implementing the diffusion process according to a particular embodiment of the invention.

5. Description of embodiments of the invention

5.1 General Principle The general principle of the invention is based on determining, head side network, a time difference between two time references obtained according to different techniques, and the transmission of a time marker directly changed in the data stream transmitted by the headend to account for this offset, or on the transmission of this time lag to the different transmitters of the network, so that the emitters change their local time reference based on this offset.

In other words, the inclusion of this time lag compensates for variations in transport time between the headend and the various transmitter sites, regardless of the broadcast standard implementation (DVB-T, DVB T2, DAB, DMB, WIMAX ...).

3 illustrates more precisely the general principle of the invention, implemented in a communication network comprising at least two separate transmitters powered by a headend.

During a first step 31, the headend receives as input on the one hand the data to be broadcast, and secondly a first time reference, obtained from an external source such as a GPS receiver. It organizes the data into frames, and inserts at least one time marker in the data stream. The thus labeled flow is then transported to the various transmission sites via a satellite link, for example. During a step 32, the headend receives the stream of data transported by the satellite. Starting of the temporal markers present in this stream, a second time reference is obtained.

The first and second time references are then compared in a step 33, to determine a time offset between the first and second time references.

Finally, during a step 34, the time difference is transmitted to different issuers, or considered to edit or cue points in the remainder of the data stream transported to the various transmitter sites. More specifically, the modification time markers allows compensate, at the headend, the influence of satellite motion, so as to emulate the reception at the emitters. It is also possible to transmit the time shift to different transmitters, which can change their local time reference based on this shift. In particular, these transmitters can be of different types, some using reliable external source for the generation of a time reference (GPS type), other regenerating the time reference from the received data stream and therefore need not using a GPS type of external reference.

This technique compensates for variations in transport time between the headend and the various transmitter sites.

It attaches below to describe various embodiments of the invention, as part of the distribution of digital terrestrial SFN type, implementing a dissemination of data in MPEG format. 5.2 Regeneration of time and frequency references As previously stated, the invention finds particular applications in broadcasting systems including issuers of various types, some using reliable external source for the generation of a time reference, and other regenerating the time reference from the received data stream. technical regeneration of temporal references are presented below 1 pps and frequency 10 MHz from the received MPEG data stream. This technique can be implemented at the network head, or at the emitters of emission sites. A) Time reference firstly We assume a time (period) transport between the headend and a site of known and constant emission, as shown in Figure 4A.

network head that is considered to what organizes data broadcasting in mégatrames considering a time reference 1 pps after a reliable source, type GPS, denoted 1 pps (GPS), and inserts at least one timestamp in the form of an MIP packet in the data stream. The flow well marked, referenced 41, is then transported to the various transmitter sites.

At the transmitting site (or headend), the transmitter receives the MPEG stream 42 which has been transported in the network. The transmitter search the MIP package_not. {.

Having found the MIP package_not. i, it is through the value "point" the first TS packet of the next mégatrame n referenced TS_notJ. The transmitter also extract the value of STS MIP packet. Having found the first TS packet_notJ, the transmitter regenerates the time reference 1 pps (rated 1 pps (MPEG)) through the STS value and the value of fixed delay, considered equal to the known and constant travel time.

In other words, the transmitter at the transmitting site or network head regenerates a time reference 1 pps by assigning the value of transport time, considered constant, fixed behind. It is thus considered that a pulse (1 pps) occurs at time defined by the reception of the first bit of the first TS packet of the next mégatrame MIP package_not. i minus the time defined by the STS value minus the fixed delay corresponding to the time of transport.

In this technique, the time reference 1 pps (MPEG) regenerated by the transmitter is synchronized with the time reference 1 pps (GPS) from a reliable source.

It is now assumed that the transport time between the headend and the transmission site is unknown or variable, which implies that the value of fixed delay is no longer equal to the time of transport. Under this assumption, as shown in Figure 4B, the time reference 1 pps (MPEG) regenerated by the issuer is not synchronized with the time reference 1 pps (GPS) from a reliable source.

There is therefore a time offset Δ between 1 pps time reference 1 pps (MPEG) and the time reference 1 pps (GPS). B) Frequency Reference The 10 MHz frequency reference is itself reconstructed from the flow of the MPEG stream. Indeed, we recall that the flow rate has been calibrated at the headend by the SFN adapter from a 10 MHz frequency reference from a trusted source, such as a GPS receiver. Specifically, the reference frequency 10 MHz is generated from the MPEG stream at the transmitter sites, thanks to a voltage controlled oscillator (VCO).

For example, Figure 5 illustrates steps of generating a reference to 10 MHz, from the data stream 42. For example, consider this stream has a flow rate at 24, 130 megabits per second. Generating a reference at 10 MHz implements the following steps:

- Serial / parallel conversion (51);

- Division (52) 36;

- Phase comparison (53) between the signal from the VCO (54) divided (55) by 110, and the data stream divided by 36 (52); - Amplification (56).

If there is no jitter ( "jitter") on the flow rate of the received MPEG stream, the 10 MHz reference frequency is very precisely the same frequency as the reference frequency 10 MHz after a GPS receiver.

So that there are exactly 10 million periods of the 10 MHz reference between two 1 pps pulses, it is also possible, alternatively, to use the principle described in Figure 4 only to "initialize" the time reference 1 pps. The latter could then be maintained using the 10 MHz frequency reference regenerated by the flow of the stream.

5.3 First embodiment is presented below, in connection with Figures 6 and 7, a first embodiment of the invention, using the principle of regeneration of time and frequency references discussed above.

According to this first embodiment, the headend includes: - satellite reception means, such as those present on emission sites (RX Network adapter), to track changes travel time related to satellite movement. These means allow to receive the MPEG stream 42 at the headend;

- A module 61 of regeneration reference 1 pps from the MPEG stream 42, denoted 1 pps (MPEG), using the technique described in paragraph 5.2;

- A GPS receiver 62, to recover a time reference 1 pps, denoted 1 pps (GPS); and

- Comparison means (64) of the two time references 1 pps (MPEG) and 1 pps (GPS). According to this first embodiment, the comparison means belong to the SFN adapter 63. The SFN adapter 63 therefore comprises, in addition to conventional inputs for data broadcast 60 on the one hand, and for a first time reference ( 1 pps (GPS)) and a first frequency reference (10 MHz (GPS)) secondly, a further input for a second time reference (1 pps (MPEG)).

These comparison means (64) for measuring, at the headend, the time variation of transport, that is to say the time offset Δ between 1 pps time reference 1 pps (MPEG) and the time reference 1 pps (GPS).

Specifically, during a first initialization phase, the data stream to be broadcast is organized in mégatrames, comprising one or more MIP packets. These MIP packets carry a time stamp synchronization STS, calculated from the time reference 1 pps coming from the GPS receiver 62. The data stream 41 thus constructed is distributed in the transport network. During a second operating phase, the transported data stream, then referenced 42, is received by the reception means of the satellite headend. The module 61 then allows to regenerate the time reference 1 pps (MPEG) from the MPEG 42.

The comparison means (64) then compares the time reference 1 pps (MPEG) and the time reference 1 pps (GPS), determining the offset Δ 1 pps time between these two references.

When the framing of the data, the SFN adapter 63 modifies the MIP packets in the STS value (65), taking into account this time lag

Δ 1 pps. In other words, changes the value of the stamp STS previously calculated from the time reference 1 pps (GPS), by adding the time offset, such that:

STS '= + Δ the STS pps.

This new value STS 'which is inserted in the MIP packets in the module 65. This time lag is taken into account to edit or cue points in the remainder of the data stream transported to the various transmitter sites.

Thus, as illustrated in Figure 7, the time reference 1 pps (MPEG) regenerated at the emission sites from the MPEG stream comprising the modified markers, using the technique described previously, is synchronized with the reference 1 pps (GPS) .

Specifically, at the transmitter sites, the first bit of the next mégatrame MIP packet stream is not broadcast by the transmitter at the time defined by the pulse 1 pps, to which value is added and STS Max delay delay, but at the moment defined by the 1 pps pulse, which the STS value added 'and the MAX_DELAY delay (or 1 pps + STS + Δ 1 pps (t) + Max delay). This variation is the same for all transmitters of the SFN, broadcast signals are well synchronized.

In particular, the comparison means (64) can regularly calculate Δlpps compensation to be made to the value STS, the following recurrence formula:

Δlpps (n + 1) = Δlpps (n) + instantaneous measured gap, where n is the index of the mégatrame.

In other words, the compensation Δ 1 pps (n + l) to bring to the mégatrame n + 1 is equal to the compensation Δ 1 pps (n) which was given to the mégatrame n, to which is added the instant measured deviation ( that is to say that the gap was still after making the compensation Δ 1 pps (n)). Note that when the measured instantaneous difference is zero, this means that the time difference between the time references Δlpps 1 pps (GPS) and 1 pps (MPEG) has been correctly calculated and that the WEPP references (GPS) and 1 pps (MPEG ), after application of the correction, are synchronized.

1 pps time reference generated from the MPEG signal is thus properly synchronized to the time reference 1 pps GPS outcome.

This solution compensates for the variation of transport time of the MPEG stream. In addition, SFN adapter works with references 1 pps and possibly 10 MHz from a reliable reception of the GPS type. Its proper functioning is assured. In particular, we note that there are exactly 10 million periods of the 10 MHz reference between two 1 pps pulse.

It is therefore possible to use, in the same broadcasting network, both transmitters using a time reference from a trusted source like GPS and transmitters using a time reference from the MPEG stream.

5.4 Second Embodiment

Is presented below in connection with Figures 8 and 9, a second embodiment of the invention. According to this embodiment, the network head comprises:

- Satellite reception means allow to receive the MPEG stream 42;

- A module 61 of regeneration reference 1 pps from MPEG 42; - A GPS receiver 62; and

- Comparison means (64) of the two time references 1 pps (MPEG) and 1 pps (GPS).

According to this second embodiment, the comparison means (64) for measuring, at the headend, the time variation of transport, i.e. the time offset Δ between 1 pps time reference 1 pps ( MPEG) and the time reference 1 pps (GPS).

This time offset is then added to the first time reference 1 pps (GPS). In other words, the network head comprises, according to this second embodiment, means 72 for predistortion of the first time reference, delivering a time reference 1 pps predistorted.

Recall that classically, the SFN adapter 71 has three entrances, one for data dissemination, and the other two for the time and frequency references.

According to this embodiment, the SFN adapter 71 is a classic adapter, comprising a first input for data dissemination 70, a second input for the 10 MHz reference frequency and a third input for the time reference 1 pps predistorted.

It is this predistorted reference that is used to change the timestamp STS in the MIP packets. Specifically, during a first initialization phase, the data stream to be broadcast is organized in mégatrames, comprising one or more MIP packets, as described for the first embodiment.

During a second operating phase, the transported data stream, then referenced 42, is received by the reception means of the satellite headend. The module 61 then allows to regenerate the time reference 1 pps (MPEG) from the MPEG 42.

The comparison means (64) then compares the time reference 1 pps (MPEG) and the time reference 1 pps (GPS), determining the time difference Δ 1 pps between these two references. The time reference 1 pps (GPS) is then changed (72) taking into account this time lag, delivering a time reference 1 pps predistorted:

1 pps predistorted = 1 pps (GPS) + Δ 1 pps

When framing data, the SFN adapter 71 determines from the time reference 1 pps predistorted value STS 'and amends MIP packets in the value STS STS '.

Thus, as illustrated in Figure 9, the time reference 1 pps (MPEG) regenerated at the emission sites from the MPEG stream comprising the modified markers, using the technique described previously, is synchronized with the reference 1 pps (GPS) .

In particular, and as shown in connection with the first embodiment, the comparison means (64) can regularly calculate Δlpps compensation to be made to reference 1 pps (GPS), by the following recurrence formula: Δlpps (n + 1) = Δlpps (n) + instantaneous measured deviation.

1 pps time reference generated from the MPEG signal is thus properly synchronized to the time reference 1 pps GPS outcome.

This second solution allows to compensate for the variation in transport time of the MPEG stream. It is therefore possible to use, in the same broadcasting network, both transmitters using a time reference from a trusted source like GPS and transmitters using a time reference from the MPEG stream.

5.5 Third Embodiment

It has now, in connection with Figure 10, a third embodiment of the invention.

According to this embodiment, the network head comprises:

- Satellite reception means allow to receive the MPEG stream 42;

- A module 61 of regeneration reference 1 pps from MPEG 42;

- A GPS receiver 62; and

- Comparison means (64) of the two time references 1 pps (MPEG) and 1 pps (GPS).

According to this third embodiment, the comparison means (64) for measuring, at the headend, the variation in transport time, that is ie the time difference Δ between 1 pps time reference 1 pps (MPEG) and the time reference 1 pps (GPS).

This time lag Δ 1 pps is used to regenerate, in a module 81, a reference frequency 10MHz predistorted. Specifically, the module 81 takes as input the reference frequency at 10 MHz coming from the GPS receiver and the time offset .DELTA.L pps predistorting the 10MHz frequency reference (GPS) taking into account the offset, and outputs a reference frequency 10MHz predistorted.

This predistorted 10MHz reference may for example be obtained by using a phase locked loop (English "Phase-Locked Loop" or

PLL) whose voltage-controlled (VCO) is controlled by the time offset

.delta.L Pps. Initially, the PLL can enslaving on the 10 MHz reference from the GPS. Then, once achieved this servo, the PLL control voltage (VCO) is increased or decreased according to the value of the time shift .DELTA.L pps to minimize it, leading to an increase or decrease in the 10MHz frequency.

Knowing that there are exactly 10 million periods of the 10MHz reference frequency between pulses 1 pps time reference 1 pps predistorted is regenerated in a module 82, from the frequency reference 1 OMHz predistorted.

The SFN adapter 83, which is a conventional adapter then comprises entries for data broadcast 80, the time reference 1 pps predistorted, and the reference frequency 10MHz predistorted.

It is this predistorted time reference which is used to change the timestamp STS in MIP packets.

More specifically, as already indicated for the other embodiments, the data stream to be broadcast is organized in mégatrames, comprising one or more MIP packets, during a first initialization phase.

During a second operating phase, the transported data stream, then referenced 42, is received by the reception means of the satellite headend. The module 61 then allows to regenerate the time reference 1 pps (MPEG) from the MPEG 42.

The comparison means (64) then compares the time reference 1 pps (MPEG) and the time reference 1 pps (GPS), determining the time difference Δ 1 pps between these two references.

The 10MHz reference frequency coming from the GPS is then locked in frequency portion of the time shift Δ 1 pps (81) delivering a reference frequency 10 MHz predistorted, and a time reference 1 pps predistorted. When framing data, the SFN adapter 83 determines from the time reference 1 pps predistorted value STS ', and changes in the MIP packets value in STS STS'.

1 pps time reference generated from the MPEG signal is thus properly synchronized to the time reference 1 pps GPS outcome. This third solution thus compensates for the variation in transport time of the MPEG stream and the frequency deviation of the reference frequency 10 MHz due to the Doppler.

In particular, it is noted that the temporal references 1 pps and frequency 10 MHz supplied to the SFN adapter 83 are connected, which means that there are exactly 10 million periods of the 10 MHz frequency reference between pulses 1 pps .

It is therefore possible to use, in the same broadcasting network, both transmitters using a time reference from a reliable source type

GPS, and transmitters using a time reference from the MPEG stream, postponing the change in the duration of transport time on the modulated signal broadcast by issuers.

Indeed, the first bit of the mégatrame following the current MIP packet is not disseminated by the issuer at the time defined by the pulse 1 pps, to which is added the STS value and Max Delay delay, but at moment defined by the 1 pps pulse, which is added value STS 'and Max delay delay (ie STS + 1 pps pps + Δ 1 (t) + MAX_DELAY).

5.6 Fourth Embodiment

Assuming we do not wish to change the time stamps in the data stream, a fourth embodiment is presented. This embodiment can synchronize the two time references, and possibly frequency, without distorting the outcome of the external reference source. As described above, this fourth embodiment makes it possible to measure the headend the time offset between a time reference provided by an external source and a time reference provided by the data stream.

According to this fourth embodiment, it directly transmits the time shift to different issuers. Thus, issuers present at the transmitter sites include, according to this embodiment, the means for obtaining this time shift and change their local time reference to reflect this shift.

The following are three embodiments.

In a first variant, illustrated in Figure 11, the network head A comprises:

- Satellite reception means, for receiving the MPEG stream 42;

- A module 61 of regeneration reference 1 pps from the MPEG stream 42, denoted 1 pps (MPEG), using the technique described in paragraph 5.2;

- A GPS receiver 62, to recover a time reference 1 pps, denoted 1 pps (GPS); and

- Comparison means (64) of the two time references 1 pps (MPEG) and 1 pps (GPS).

In this first variation, the time offset between the time reference 1 pps (MPEG) and 1 pps (GPS), resulting from the comparison means 64 is transmitted to at least one emission site B, using means external transmission 111.

This means of external transmission 111 is for example a GPRS or IP link transmission.

At the emission site B, 42 received MPEG stream is used to regenerate the reference 1 pps, using the technique described in Section 5.2.

The time shift obtained at the transmission site is added (112) to signal 1 pps (MPEG) regenerated at the transmitting site, corresponding to the local time reference. This gives a time reference 1 pps compensated. In other words, the time reference 1 pps (MPEG) finds himself aligned with an external time reference (eg 1 pps (GPS)), compensating the time difference Δ 1 pps.

For this embodiment works effectively, it is desirable to transmit several times a day this time offset value to each of the broadcasting sites. In a second variant, illustrated in Figure 12, the network head A includes, in addition to the elements mentioned in connection with the first embodiment, means 121 for modifying the data stream.

These means 121 are used to insert in the data stream to be broadcast, the time offset resulting from the comparison means 64. More specifically, the time offset is inserted into a TS packet having a PID that is to say a TS packet identifier, dedicated. For example, this time lag is inserted in replacing a padding packet of the stream by a packet identified by a PID phantom, that is to say a packet with a PID that is not described in the tables and which do is not reserved by the standard. Alternatively, this time lag is inserted in replacing a padding packet of the stream by a signaling packet which does not alter the nature of the signal to be broadcast.

We consider this variant as the issuer knows the specific package labeling with the time difference (PID ghosts or packages that signaling does not change the nature of the signal to be broadcast). It is also possible to insert several time shift values in the flow by introducing specific packet marking in each of mégatrames.

This insertion step is for example implemented by a slightly modified MIP inserter. At the emission site B, the received MPEG stream is used to regenerate the reference 1 pps, using the technique described in Section 5.2.

A retrieval module 122 is also provided for extracting, in the specific packet marking, the time difference measured at the headend. Extract the time offset is then added (123) to the 1 pps signal (MPEG) regenerated at the transmitting site, delivering a time reference 1 pps compensated.

The time reference 1 pps (MPEG) finds himself aligned with the time reference 1 pps (GPS), compensating the time difference Δ 1 pps. It may also be noted that the TS packet dedicated marking is treated transparently by the modulators (for example referenced 102 and 112 in relation to Figure 1) at the emission sites B.

According to a third variant, it is also possible to insert one or more marker (s) Time (s) modified (s) as described in connection with the first three embodiments in one or more specific TS packets to the label. For example, these specific package labeling are identified by a PID ghost or a packet signaling does not change the nature of the signal to be broadcast.

This creates specific packet marking (MIP ghosts for example), with for example a value STS '(STS' = STS + Δ 1 pps), coexisting in the data stream with MIP packets "classic", with a value STS.

This insertion step is for example implemented by a slightly modified MIP inserter, allowing the introduction of specific marking packets in the MPEG stream. The data flow transported from the headend to the various issuers therefore has a double marking MIP:

- Marking MIP "classic", used by the modulators (eg, referenced 102 and 112 in relation to Figure 1); - The specific MIP marking transparent for modulators, and used only when synchronizing system in emission centers for generating a reference 1 pps aligned reference 1 pps after a reliable source (GPS).

For example, it is possible to indicate to the transmitter (eg the timing equipment) PID MIP ghosts packets it uses to regenerate his reference 1 pps.

It is also possible to insert, at the level of emission centers, between the TS stream from the satellite reception and the 'regenerator WEPPA - 10 MHz ", a module which reverses the PID conventional MIP and MIP packets ghosts packages .

In particular, the specific packet marking inserted into the stream before transmission can be filtered before extraction time markers MIP "classic", that is to say at each transmitter, for example by crushing this or these specific package labeling with a padding packet.

It may be noted that filter specific packet marking is deterministic. Thus, as part of an SFN network type, it does not change the appearance of the SFN broadcast signal.

Thus, the time reference 1 pps (MPEG) regenerated from the data stream will be determined from time markers bearing the STS value '. Thus, the time reference 1 pps (MPEG) can be found aligned with an external time reference (e.g., 1 pps (GPS)). 5.7 Structure of the headend

It has now, in connection with Figure 13, the simplified structure of a network head by implementing a diffusion process according to one of the particular embodiments described above.

Such headend includes a memory 131, a processing unit 132, equipped for example with a .mu.P microprocessor and driven by the computer program 133, implementing the broadcasting method of the invention.

At initialization, the computer program code instructions 133 are for example loaded into a RAM before being executed by the processor of the processing unit 132. The processing unit 132 receives as input data disseminate D, and a time reference 1 pps (and possibly frequency 10 MHz). The microprocessor of the processing unit 132 implements the steps of the diffusion process described above, according to the instructions of the computer program 133.

For this, the network head comprises, besides the memory 131, means for obtaining a first time reference from an external source, means for obtaining a second time reference from said flow data received by the network head, means for comparing said first and second time references, and means for transmitting said time offset or at least one of said time markers changed in accordance with said temporal offset. These means are driven by the microprocessor of the processing unit 132.

The processing unit 132 outputs the time difference or the data stream comprising at least one modified time marker.