WO2009016175A1

WO2009016175A1 - Method and device for synchronising a data flow in a single frequency network

Info

Publication number: WO2009016175A1
Application number: PCT/EP2008/059932
Authority: WO
Inventors: Antoine Clerget; Luc Ottavj
Original assignee: Udcast
Priority date: 2007-08-02
Filing date: 2008-07-29
Publication date: 2009-02-05
Also published as: FR2919775A1; EP2179523A1

Abstract

According to a first aspect, the invention relates to a method for adapting a data flow for the isochronous distribution thereof by a transmitter network operating at a single frequency, that comprises: cutting the data flow into a plurality of megaframes and inserting into each megaframe an initialisation packet to be used by the transmitters for synchronising the distribution of the data flow, the initialisation packet of a megaframe including a field STS for indicating the beginning of the transmission of the next megaframe to the transmitters; evaluating for each megaframe the beginning of the transmission of the next megaframe relative to a time reference not in high resolution and including an imprecision; characterised in that it comprises for each megaframe applying a correction to the evaluation of the beginning of the transmission of the next megaframe for delaying said evaluation of the imprecision of the time reference, and filling the STS and maximum_delay fields of the megaframe while taking into account the corrected evaluation. The invention also relates to a device including means for implementing the method.

Description

METHOD AND DEVICE FOR SYNCHRONIZING A DATA STREAM IN A SINGLE FREQUENCY NETWORK

The field of the invention is that of communication systems, and more specifically that of communication systems for broadcasting multimedia content to fixed or portable receivers via digital radio networks of the single frequency network type (network SFN for "Single Frequency Network").

The DVB-T, DVB-H, DVB-SH standards are examples of digital terrestrial broadcasting standards for broadcasting content to fixed or portable receivers in which broadcast streams are synchronized for distribution in a network environment. single frequency.

In an SFN network, a content source transmits the data to be broadcast to a plurality of transmitters synchronized with each other. Synchronization is indeed necessary so that all the transmitters of the SFN network transmit the same binary information (bits) in the same frequencies and at the same time.

In order for all the transmitters to be able to broadcast the transport stream synchronously, the transport stream includes markers that make it possible to define a reference between a position in the bit stream and a time reference.

In the case of DVB-T and also DVB-H, a technique for synchronizing transmitters in a single frequency network is specified in ETSI TS 101 191 entitled "Digital Video Broadcasting".

(DVD); DVB mega-frame for Single Frequency Network (SFN) Synchronization.

This specification proposes that an SFN adapter with the common transport stream facilitates the synchronization of the transmitters by transmitting to the transmitters temporal information which is computed on the basis of a repetitive time reference and a frequency reference. references being obtained from a high-resolution time source such as a GPS satellite positioning system. This time reference is broken down into a 10 MHz clock signal giving an accuracy of 100 ns and a pps signal ("pulses per second") giving the same time the division of time in seconds.

The SFN adapter specified in ETSI TS 101 191 is more specifically responsible for forming groups of temporally marked frames (called mega-frames, "mega-frame" in the English terminology) and for sending these mega-frames. on the distribution network to the transmitters of the SFN network, thus allowing the modulators associated with the transmitters to be synchronized accurately and to deliver the same bits on the same COFDM carriers at the same time. All the signals broadcast by the transmitters are thus identical, which makes it possible to achieve simultaneity of reception of the information by all the users. A mega-frame consists more precisely of a group of n packets

TS or RS packets (i.e. TS packets to which Reed Solomon OFDM coding information has been added), where n is an integer that depends on the number of RS packets per mega-frame OFDM in the mode DVBT transmission used. A mega-frame contains a synchronization marker known as MIP ("Mega-frame Initialization Packet"). An MIP is an MPEG-2 packet with a dedicated identifier (PID). The MIP packet of a mega-frame of index M mainly makes it possible to determine the starting point of the next mega-frame of index M + 1 and to provide the emitters the moment at which they must start the transmission.

The MIP notably carries information intended for the transmitters and concerning the position of the first packet of the next mega-frame in the transport stream (expressed in terms of the number of MPEG-2 TS packets), timing information (with a precision of 100 ns) indicating to the modulator when the modulation of the next mega-frame must be started, as well as configuration values for modulation and transmission parameters.

The time information included in the MIP packet of a megatram is mainly the "STS" field (Synchronization Time-Stamp) and the "maximum_delay" field, as defined in the ETSI specification TS 101 191 and illustrated in its appendix B " Functional description of SFN synchronization ").

The STS field of the MIP packet of an M-mega-frame contains the time difference between the last pulse of the reference "1 pps" preceding the start of the M + 1 index mega-frame and the true departure. this mega-frame at the SFN adapter to the transmitters.

The "maximum_delay" field contains the temporal difference between the moment of the synchronous transmission of the start of the mega-frame M + 1 by the antennas of all the transmitters in the cell of the network.

SFN, and the departure of the mega-frame M + 1 at the SFN adapter to the transmitters. The value of the "maximum_delay" field must be greater than the sum of the longest delay in the distribution network and the delays in modulators, transmitters and antennas. The maximum value of the "maximum_delay" field is one second.

A transmitter receiving the start of a mega-frame at T _rec (expressed in unit of 100 ns relative to the current pulsation of the reference 1 pps) will delay this mega-frame of a duration T _d eiay which is equal at (STS + maximum_delay - T _rec ) modulo 10 ⁷ .

As we have seen, state-of-the-art SFN adapters are provided with a time reference provided by a high-resolution time source (for example via a GPS), and realized on computer systems or real-time electronics. However, it would be preferable to perform the SFN adaptation function at a lower cost, for example on non-real-time computer systems with a less accurate time reference than GPS.

In this respect, it has been proposed by document WO 2006/046107 that a non-high-resolution "imprecise" temporal reference be delivered to the SFN adapter. More specifically, this document provides that the SFN adapter receives the time reference of a server complying with the Network Time Protocol (NTP). In this way, the STS field of the MIP is determined with respect to the imprecise time reference. In the technique proposed by this document, the transmitters of the single frequency network implement the synchronization recommended in the ETSI specification, which consists for a transmitter to compare the STS field (here inaccurate because established with respect to the imprecise time reference of the server NTP) with the arrival time T _rec of the mega-frame via the distribution network (this arrival time T _rec being precise since it is determined with respect to a precise temporal reference delivered to the transmitters by a high-resolution source of type GPS). The transmitter then decides to delay the emission of the mega-frame by a delay T _d eiay given by the difference between the maximum delay to reach the transmitter furthest distant and the difference between the arrival time T _rec of the mega-frame and the STS. The transmitter thus retards the emission of T _from iay = (STS + maximum_delay - T _rec ) modulo 10 ⁷ .

Although the method recommended in this document makes it possible to overcome the use of a high-resolution temporal reference (for example a GPS source) at the level of the SFN adapter, it nevertheless has drawbacks.

In particular, when the delay to reach the nearest transmitter via the distribution network is less than the inaccuracy of the non-high-resolution time source delivered to the SFN adapter, it is possible that the synchronization is not correctly performed.

Consider the following case in which: the imprecise time delivered to the SFN adapter is greater than the absolute time of a quantity Δ. For example, the SFN adapter thinks that it has sent a mega-frame at pps + 200, while the absolute sending time of this mega-frame is actually pps + 100; - the delay to reach the nearest transmitter is less than Δ.

Suppose that the transmission time on the distribution network to reach the nearest transmitter from the SFN adapter is 50. This transmitter will receive at absolute time (pps + 100) +50, information that is indicates as having been sent to pps + 200. This issuer will therefore deduce that the information is part of the previous pps, and that it therefore took 950 ns to reach him.

This transmitter can then delay the transmission only if the "maximum_delay" field in the MIP packet is greater than 950 and all the repeaters are actually less than 50 of the SFN adapter, which excludes use in such a case the satellite or the internet as a distribution network. These media actually have delivery times well above 50ns, which requires the operator to indicate delays of the order of 950ns when in reality these delays are of the order of one hundred ns. On the basis of this observation, the aim of the invention is to propose a technique for the synchronization of transmitters in a single frequency network which does not require the use of a high-resolution time reference (of the GPS source type) at the level of the SFN adapter while nevertheless allowing the implementation of the standard synchronization, in accordance with the ETSI specification, at the level of the transmitters, and which moreover avoids the disadvantages of the technique proposed in the document WO 2006/046107.

For this purpose, according to a first aspect, the invention proposes a method of adapting a data stream for isochronous distribution by a network of transmitters operating on a single frequency, comprising: splitting the data stream into a plurality of mega-frames, and inserting in each mega-frame an initialization packet for use by the transmitters to synchronize the distribution of the data stream, the packet of initialization of a mega-frame comprising an STS field intended to indicate the start of the transmission of the next mega-band to the transmitters,

the evaluation, for each mega-frame, of the beginning of the transmission of the next mega-frame with respect to a non-high resolution temporal reference having an inaccuracy, characterized in that, for each mega-frame, a correcting the evaluation of the start of transmission of the next mega-frame to reduce said evaluation of the inaccuracy of the time reference, and informing the STS field of the mega-frame taking into account the corrected evaluation. Some preferred, but not limiting, aspects of this method are as follows:

the evaluation, for each mega-frame, of the beginning of the transmission of the following mega-frame to the transmitters comprises: the determination, relative to the non-high-resolution time reference, of the instant t _M ip to which the insertion of the initialization packet in the mega-frame is supposed to be carried out, and o the addition of the processing and transmission time t _p of the data of the mega-frame remaining to flow until the beginning of the transmission from the mega-frame following said time t _M ip to which the insertion is supposed to be carried out, so that the transmission start of the next mega-burst to the transmitters is evaluated at tmip + tp, and in which the field STS is given taking into account the corrected evaluation according to STS = t _M ip + tp - dt. the initialization packet of a mega-frame further comprises a maximum_delay field representing the difference between the beginning of the transmission of a mega-frame to the transmitters and the start Ttransmitted of the synchronous distribution of this mega-frame by the transmitters, wherein the data stream is intended to be transmitted to the transmitters via a distribution network having a delay of maximum propagation D _MAX , and in which one informs the field maximum_delay of each mega-frames according to maximum_delay = D _M AX + 2 ^* dt.

the data stream being transmitted to the transmitters via a distribution network having a minimum propagation delay D _min and a maximum propagation delay D _MAX , a virtual distribution network with a minimum propagation delay of zero and a maximum propagation delay are considered; equal to D _MAX - D _mm and the STS field of a mega-frame is informed by taking into account the corrected evaluation of the start of transmission of the next mega-frame on said virtual distribution network;

the evaluation, for each mega-frame, of the beginning of the transmission of the next mega-frame on the virtual distribution network comprises: the determination, with respect to the non-high-resolution time reference, of the instant t _M ip which is supposed to be carried out the insertion of the initialization packet in the mega-frame, and o the addition of the processing time and transmission t _p data of the mega-frame remaining to flow until the beginning of transmitting the next mega-frame to the transmitters at said instant _M ip to which the insertion is supposed to be performed, and the minimum propagation delay D _mn , so that the transmission start of the next mega-frame on the virtual distribution network is evaluated at tmip + tp + Dmin, and the STS field is populated taking into account the corrected evaluation according to STS = tmip + tp + D _mn -dt. the initialization packet of a mega-frame further comprising a maximum_delay field representing the difference between the beginning of the transmission of a mega-frame to the transmitters and the Ttransmitted start of the isochronous distribution of this mega-frame by the transmitters, the maximum_delay field of the megatrames is given according to maximum_delay = D _M AX-D _mιn + 2 ^* dt.

the non-high-resolution time reference is provided by a computer system having a clock whose drift is stabilized in accordance with the Network Time Protocol (NTP). According to a second aspect, the invention proposes a device for adapting a data stream with a view to its isochronous distribution by a network of transmitters operating on a single frequency, comprising:

means for splitting the data stream into a plurality of mega-frames adapted to insert in each mega-frame an initialization packet (MIP) intended to be used by the transmitters to synchronize the distribution of the data stream; a mega-frame initialization packet having a field (STS) for indicating the start of transmission of the next mega-frame,

evaluation means, for each mega-frame, of the beginning of the transmission of the next mega-frame with respect to a non-high-resolution temporal reference having an inaccuracy, characterized in that the means for splitting the stream are in further adapted to correct said evaluation of the start of transmission of the next mega-frame by delaying the inaccuracy of the time reference and to inform the STS field of a mega-frame taking into account said corrected evaluation.

Other aspects, objects and advantages of the present invention will appear better on reading the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and with reference to the appended drawing in which: Figure 1 schematically shows the cutting of the data stream into a plurality of mega-frames. The invention relates to a method for adapting a data stream with a view to its isochronous distribution by a network of transmitters operating on a single frequency, and a device capable of implementing this method.

The data stream may be of the DVB-H compatible type. It is mentioned here that, without this being for all that limiting, we will take the example of a DVB-H stream in the following description.

The data stream can also be a DVB-SH compatible stream. It will be remembered that in DVB-SH, the isochrony of the transmitters is ensured in the same way as in DVB-H, except that: - a "mega-frame" carries in DVB-SH the name of "SH-frame";

- that the MIP package carries the name of SHIP ("SH Initialization Packet") in DVB-SH, the SHIP packet including, just like the MIP packet in DVB-H, the STS and maximum_delay fields. The mega-frame expression that will be used subsequently must not be understood in a limiting manner; this expression also having particular vocation to apply to the "SH-frame" of a DVB-SH compatible stream.

In the same way, the MIP expression that will be used later must not be understood in a limiting manner; this expression also having particular application to the SHIP packet of a DVB-SH compatible stream.

The device can take the form of an SFN adapter typically placed in each cell behind an IP encapsulator or multiplexer. The device can however be integrated directly into the IP encapsulator or into the multiplexer. It will be understood that the method according to the invention can be implemented either in an SFN adapter, in an IP encapsulator or in a multiplexer fulfilling the SFN adaptation functions. However, we will take in the following, without however being limiting, the classic example of a standalone SFN adapter. In the context of the invention, it is proposed that the SFN adapter not benefit from a high resolution time source as it is recommended by ETSI specification 101 191 (for example a time source delivered by a GPS receiver), but a non-high resolution time source characterized by imprecision dt.

Thus, where classically, the instant at _M ip at which the insertion of the MIP packet in a mega-frame is performed is particularly precisely determined (this instant being evaluated with respect to the time reference of the

GPS), in the context of the invention we can only evaluate this moment with an inaccuracy corresponding to that of the time reference used.

The moment at which the insertion of the MIP packet is supposed to be carried out is thus evaluated at t _M ip in the context of the invention, with respect to the non-high resolution time reference, with at _M ip - dt <t _M ip < has _M ip + dt

The invention pays attention to the fact that the STS field of a mega-frame does not indicate a start of transmission of the next mega-frame (calculated for example from the imprecise time t _M ip insertion of the MIP packet) which is greater than the real beginning of transmission (linked to the absolute time at _M ip) of this next mega-frame.

In this regard, the invention proposes to apply, for each mega-frame, a correction to the evaluation of the beginning of the transmission of the next mega-frame to reduce said evaluation of the inaccuracy of the time reference. We then inform the STS field of the mega-frame taking into account the corrected evaluation.

FIG. 1 shows the structure of a stream of data divided into a plurality of mega-frames, in particular the mega-frames of index M and M + 1. The MIP packet is inserted into the mega-frame of index M, and contains the STS field which represents the time difference between the last pulse of the reference 1 pps which precedes the beginning of the next megatram of index M + 1 and the true departure of this mega-frame at the SFN adapter for issuers. In the example here selected, the MIP packet is more precisely positioned in the mega-frame of index M so that p packets of this mega-frame are interposed between the MIP packet and the beginning of the next mega-frame of M + 1 index. When the absolute time at _M ip at which the insertion of the MIP packet of the mega-frame of index M is performed with respect to the high-resolution time reference type GPS, the field STS can be calculated as being equal to ( at _M ip + tp) modulo 10 ⁷ , where tp represents the processing and transmission time of the p packets of the mega-frame of index M to be transmitted before the start of the next mega-frame of index M + 1.

In the context of the invention, this absolute time at _M ip is not known, and only an inaccurate evaluation t _M ip of this absolute time is available.

However, because of the inaccuracy of this evaluation, it is possible in particular that the evaluation of the transmission start of the next mega-frame is greater than the true transmission date of the next mega-frame (tmip + tp> atmip + tp).

The invention then proposes to apply a correction to the evaluation of the start of the transmission of the next mega-frame M + 1, by decreasing the said evaluation by an amount equal to the inaccuracy dt of the non-high time reference. resolution, and fill in the STS field of the packet

MIP of the mega-frame M taking into account said corrected evaluation.

In particular, according to a first possible embodiment of the invention, and when the insertion time of the MIP packet is evaluated at t _M ip, the invention proposes to inform the STS field according to STS = t _M ip + tp - dt .

Furthermore, considering that the distribution network of the data stream intended for the transmitters of the SFN network has a maximum propagation delay D _MAX , the field "maximum_delay" of each of the mega-frames is given as maximum_delay = D _MAX + 2 ^* dt. Thus, each transmitter receiving the mega-frame M index + 1 to T _rec, conventional manner retards the diffusion of this T mega-frame _of eiay = {STS + maximum_delay - T _rec} modulus 10 ⁷ be when the time departure of the mega-frame M + 1 is evaluated with respect to the evaluation of the insertion time of the MIP packet in the mega-frame M, T _of iay = {t _M ip + tp + D _M AX + dt - T _rec. modulo 10 ⁷ . It will be noted that the field "maxinnunn_delay" must be less than 1 second, the implementation of this first mode of the invention requires that the longest path on the distribution network is less than (1s - 2 * dt).

This constraint is, however, quite slight insofar as the inaccuracy values dt commonly obtained with conventional means, such as for example a Unix system having an NTP synchronization, are of the order of ten milliseconds, very low compared to the total delay of 1 s.

This constraint can moreover be minimized by a second possible embodiment of the invention. According to this mode, assuming that the distribution network has a propagation delay varying between a minimum propagation delay D _mn and a maximum propagation delay D _MAX , we choose to consider a virtual distribution network characterized by a minimum propagation delay zero and a maximum propagation delay equal to D _MAX - D _min . In other words, we consider that the SFN adapter is virtually placed at the nearest transmitter adapter (D _mιn the SFN adapter into the true distribution network) and the other issuers (to Di from the SFN adapter in the real distribution network) are brought back to Di-D _mιn from the virtual SFN adapter. In the context of this second embodiment, the STS field of a mega-frame is informed by taking into account the corrected evaluation of the start of transmission of the next mega-frame on said virtual distribution network.

Considering the case where the evaluation of the beginning of transmission of the next mega-frame is carried out on the basis of the evaluation at t _M ip of the insertion time of the MIP packet, one informs the field STS of a mega-frame taking into account the corrected evaluation according to STS = tmip + tp + D _min - dt.

In the context of this second embodiment of the invention, the "maximum_delay" field of the MIP packet of the mega-frames is advantageously given according to maximum_delay = D _M AX-D _mιn + 2 ^* dt. This is advantageous in that the constraint that the maximum delay of the distribution network must be less than 1 second is reduced. Indeed, this constraint of one second is no longer linked to the path of the longest to reach a transmitter, but is mainly related to the difference in time between the longest path and the shortest path.

This embodiment therefore allows the distribution of transport streams

SFN on networks having more than one second of propagation delay, and more particularly the use in such networks of equipment introducing a high latency, insofar as this latency is between a minimum and a maximum. In such cases, only the difference between the maximum and minimum latencies will be taken into account in the allocated budget of 1 s.

Moreover, the validity of the data stream is not altered since each transmitter receiving the M index mega-frame T + 1 _rec, delayed conventionally dissemination of this T _mega-frame IAY = {STS + maximum_delay - T _rec } modulo 10 ^{7 that} is T _d eiay = {tMip + tp + Dmin - dt + DMAX -Dmin + 2 ^* dt - T _rec } modulo 10 ⁷ or Tdelay = {t _M ip + tp + D _M AX + dt - T _rec } as is the case for the first embodiment of the invention. Returning to the general description of the invention, consider the case where the MIP packets are always inserted at the same position in the megatrams (so that p packets follow the MIP packet and Pp packets precede the MIP packet), the STS fields. (i) and maximum_delay (i) of the MIP packet of the mega-frame of index i can be obtained from the corresponding fields of the MIP packet of the mega-frame of index i-1 according to:

STS (i) = STS (i-1) + P ^* T, where T represents the transmission processing time of a packet of a mega-frame;

- maximum_delay (i) = maximum_delay (i-1).

The non-high resolution temporal reference used in the context of the invention is understood as a reference having the resolution recommended in the ETSI 101 specification 191. These include a reference provided to the SFN adapter by a non-real-time computer system, for example having a clock whose drift is stabilized in accordance with the NTP protocol (Network Time Protocol). The resolution of the non-high-resolution time reference is for example from 1 ms to several tens of ms (where that of a GPS is 100ns).

The following is an example given in the introduction according to which:

the imprecise time delivered to the SFN adapter is greater than the absolute time of a quantity Δ (the inaccuracy dt therefore being greater than or equal to Δ). For example, the SFN adapter thinks that it has sent a mega-frame at pps + 200, whereas the absolute sending time of this mega-frame is actually pps + 100 (here the inaccuracy dt is greater than or equal to 100).

- the delay to reach the nearest transmitter is less than Δ. Suppose that the transmission time D _mn on the distribution network to reach the nearest transmitter from the SFN adapter is 50. This transmitter will then receive at an absolute time (pps + 100) +50, information that it is indicated, via the STS field of the previous mega-frame, as having been sent to:

- (pps + 200) -dt in the context of the first embodiment of the invention, that is to say an information that is indicated as having at the latest been sent to pps + 100; The receiver will consider in this case that the mega-frame has 50 ns to reach him, which corresponds to the actual delay on the network;

- (pps + 200) -dt + D _mιn within the second possible embodiment of the invention, that is to say, information which indicates it to have later been sent to 150 pps + . The receiver will consider in this case that the mega-frame has 0ms to reach him. It can be seen that in both cases the transmitter will be able to apply standard SFN mechanisms to calculate the delay in transmitting the mega-frame, and ensure the isochronous operation of the network. Note also that the disadvantage reported in the example presented in the introduction (950ms delay and sending assumed the previous pps) is this time avoided.

As already explained above, the invention is not limited to a method but also extends to a device for adapting a data stream with a view to its synchronous distribution by a network of transmitters operating on a single frequency , comprising means configured to implement the method, and in particular:

evaluation means, for each mega-frame, of the beginning of the transmission of the next mega-frame with respect to a non-high-resolution time reference having an inaccuracy,

means for splitting the data stream into a plurality of mega-frames, adapted to insert in each mega-frame an initialization packet (MIP), adapted to correct said evaluation of the beginning of transmission of the next mega-frame by delaying the inaccuracy of the time reference and to fill the STS field of the MIP packet taking into account said corrected evaluation.

Claims

A method of adapting a data stream for isochronous distribution by a network of transmitters operating on a single frequency, comprising:

splitting the data stream into a plurality of mega-frames, and inserting in each mega-frame an initialization packet for use by the transmitters to synchronize the distribution of the data stream, the packet of initialization of a mega-frame comprising an STS field intended to indicate the start of the transmission of the next mega-band to the transmitters,

the evaluation, for each mega-frame, of the beginning of the transmission of the next mega-frame with respect to a non-high resolution temporal reference having an inaccuracy, characterized in that, for each mega-frame, a correcting the evaluation of the start of transmission of the next mega-frame to reduce said evaluation of the inaccuracy of the time reference, and informing the STS field of the mega-frame taking into account the corrected evaluation.

The method of claim 1, wherein the evaluation, for each mega-frame, of the start of transmission of the next mega-frame to the transmitters comprises:

determining, with respect to the non-high-resolution time reference, the time (t _M ip) at which the insertion of the initialization packet (MIP) is supposed to be performed in the mega-frame, and

the addition of the processing and transmission time (t _p ) of the data of the mega-frame remaining to flow until the beginning of the transmission of the next mega-frame at said instant (t _M ip) which is supposed to be performed inserting, so that the transmission start of the next mega-frame to the transmitters is evaluated at tmip + tp, and wherein the STS field is populated taking into account the corrected evaluation according to STS = t _M ip + tp - dt.

3. Method according to one of claims 1 to 2, wherein the initialization packet of a mega-frame further comprises a maximum_delay field representing the difference between the start of the transmission of a mega-frame to destination. transmitters and the start (Tt _ra nsmιtted) of the isochronous distribution of this mega-frame by the transmitters, wherein the data stream is intended to be transmitted to the transmitters via a distribution network having a maximum propagation delay D _MAX , and in which one informs the field maximum_delay of each mega-frames according to maximum_delay = D _MAX + 2 ^* dt.

The method according to claim 1, wherein the data stream is transmitted to the transmitters via a distribution network having a minimum propagation delay D _mn and a maximum propagation delay D _MAX , in which a virtual distribution network of minimum propagation delay of zero and a maximum propagation delay equal to DMAX-Dmm and in which the STS field of a mega-frame is entered, taking into account the corrected evaluation of the start of transmission of the next mega-frame on said virtual distribution network.

The method of claim 4, wherein the evaluation, for each mega-frame, of the start of transmission of the next mega-frame on the virtual distribution network comprises:

the addition o of the processing and transmission time (t _p ) of the data of the mega-frame remaining to flow until the beginning of the transmitting the next mega-frame to the transmitters at the instant (t _M ip) to which the insertion is to be performed, o and the minimum propagation delay D _mm , so that the transmission start of the next mega-burst on the virtual distribution network is evaluated at tmip + tp + Dmin, and in which the STS field is populated taking into account the corrected evaluation according to STS = tmip + tp + D _mn -dt.

The method of claim 5, wherein the mega-frame initialization packet further comprises a maximum_delay field representing the gap between the start of transmission of a mega-frame to the transmitters and the beginning. (Ttransmitted) of the isochronous distribution of this mega-frame by the emitters, and in which one informs the field maximum_delay of the mega-frames according to maximum_delay = D _M AX - Dmin + 2 ^* dt.

7. Method according to one of the preceding claims, wherein the non-high-resolution time reference is provided by a computer system having a clock whose drift is stabilized in accordance with the NTP protocol (Network Time Protocol).

A device for adapting a data stream for isochronous distribution by a network of transmitters operating on a single frequency, comprising:

means for splitting the data stream into a plurality of mega-frames adapted to insert in each mega-frame an initialization packet (MIP) intended to be used by the transmitters to synchronize the distribution of the data stream; a mega-frame initialization packet having a field (STS) for indicating the start of transmission of the next mega-frame, evaluation means, for each mega-frame, of the beginning of the transmission of the next mega-frame with respect to a non-high-resolution temporal reference having an inaccuracy, characterized in that the means for splitting the stream are in further adapted to correct said evaluation of the start of transmission of the next mega-frame by delaying the inaccuracy of the time reference and to inform the STS field of a mega-frame taking into account said corrected evaluation.

9. Device according to the preceding claim, incorporating a clock providing said non-high resolution time reference.