WO2009007261A2

WO2009007261A2 - Method and device for transmitting bursts of variable size

Info

Publication number: WO2009007261A2
Application number: PCT/EP2008/058311
Authority: WO
Inventors: Antoine Clerget; Patrick Cipiere
Original assignee: Udcast
Priority date: 2007-07-06
Filing date: 2008-06-27
Publication date: 2009-01-15
Also published as: WO2009007261A3; FR2918531B1; FR2918531A1; EP2174433A2

Abstract

The invention relates to a method, and a device configured to implement the method, of transmitting a stream of data comprising a plurality of individual streams (ES) subdivided in time to be transmitted in the form of bursts, at least one time indicator (delta-t) indicating in a burst of an individual stream the starting time of the next burst of the same individual stream, the method comprising steps according to which: at the latest on transmitting a burst (Bi,j), a prediction is made as to the starting time (ti,j,+1) of the next burst (Bi,j+1) of the same individual stream (ESi), and the time indicator of the burst (Bi,J) is entered as a function of the starting time predicted for the next burst (Bi,j+1), at the predicted starting time (ti,j+1), the start of the transmission of the next burst is forced, and if a burst (Bk,j) of another individual stream (ESk) is then currently being transmitted, said next burst (Bi,j,+1) is transmitted multiplexed with said burst (BkJ) of another individual stream (ESk) currently being transmitted.

Description

METHOD AND DEVICE FOR SENDING SALVES OF SIZE

VARIABLE

The field of the invention is that of communication systems, and more specifically that of communication systems for the transmission of multimedia content to portable terminals via digital broadcasting networks.

The DVB-H standard is an example of a digital terrestrial broadcasting standard for the transmission of content to portable terminals, in which medium speed audio / video IP streams of the order of

300-500kbs are transmitted on radio channels whose flow is greater by an order of magnitude (5-20Mbs).

Instead of sending data streams continuously by multiplexing on the channel the various flows at the IP or MPEG2 level, which would require for the terminal the permanent listening of the channel, the streams are sent by bursts of limited duration (typically 100ms ) using the entire bandwidth allocated to DVB-H on the channel.

Thus on a channel on which 10Mbs are devoted to DVB-H, the flow of a service (or elementary flow) average flow 500kbs can be achieved by repeating every 2 seconds bursts of 100ms. The terminal can then simply listen to the channel intermittently (100 ms every 2 seconds), which saves 95% of the power consumption of its radio circuit.

Streams broadcast in DVB-H, and more particularly video streams, are characterized by a variable bit rate which originates from the encoding techniques. Indeed, at each change of scene, a main image of large size is sent. This main image is followed by images containing the differences with respect to the main image, and which therefore have a smaller size. The optimal support of these variable-rate streams in a burst transmission context is an important technical and economic issue for the development of Mobile TV services.

Constitution of bursts

The burst transmission mechanism is characterized by a signaling that allows the terminal to announce the times when it will have to be listening to receive bursts. The implementation of this signaling is described in particular in document ETSI EN 301 192: "Digital Video Broadcasting (DVB); DVB specification for data broadcasting ", in particular paragraph 9:" Time slicing and MPE-FEC ".

From a practical point of view IP datagrams are encapsulated in MPE (Multi-Protocol Encapsulation) sections, which are themselves transported in MPEG-2 packets.

In the header of the MPE sections of a burst of a service, there is a field (delta-t) indicating the time at which the first MPE section of the next burst will appear for the same service. In addition, a salvo is characterized by:

- its duration: Bd in seconds;

- the bit rate used for its transmission: Bb ("Burst Bandwidth" in b / s);

- the time between the end of the salvo and the beginning of the next salvo for the same service: Ot (Off Time). We deduce the repetition interval Ri = Bd + Ot, as well as the number of bits constituting the burst Bs = Bb ^* Bd.

The number of services (or elementary streams) passed at a given moment is known.

The bursts of a service can be of fixed size (identical or different from one service to another) or variable, which will pose in a different way the problem of the calculation of the delta-t fields. Fixed size

When the bursts are of fixed size, it is easy to calculate the delta-t of each MPE section from the durations of the various services remaining to be sold, the number of MPEG-2 packets already transmitted in the burst and the dedicated flow on the channel to the transmission of bursts.

This calculation "a priori" can be done during the constitution of the salvo, without introducing additional delay in sending the data.

The use of fixed size bursts will also facilitate the hand-over mechanisms for which the same contents are available from one cell to another in different frequencies. Among these mechanisms, the so-called "phase shift", based on the impossibility of rendering isochronous the arrival and the processing of IP datagrams in the various encapsulators IPE ("IP Encapsulator"), consists in ordering differently from one IPE to another the bursts of the various services so as to allow a changing cell terminal to avoid losing IP datagrams and to find in the new cell some of the IP datagrams already received. These mechanisms can easily be implemented when the bursts have a fixed size. Variable size When the bursts have a variable size, the prior knowledge of the sizes of bursts remaining until the occurrence of the next burst for the same service is impossible. Calculation "a priori" delta-t is therefore also impossible.

One method commonly used in the state of the art is to delay data forgiveness to obtain a complete view of the bursts between the current burst and the next salvo of the same service. A "posterior" calculation of the delta-t is then performed.

This method has the disadvantage of introducing a delay in the transmission, this delay can reach a maximum of 39 seconds. However, the introduction of significant delays is not desirable, especially when live events are broadcast. As we have seen previously, most of the streams broadcast in DVB-H are video streams whose bit rate is variable by nature. These flows are also designated by the acronym VBR of the English expression "Variable Bit Rate". Such VBR streams can be passed with a fixed size bursts policy. In such a case, the size of the bursts will have to be wedged on the maximum which will cause a waste of bandwidth. This waste, estimated by the current art at 30-40%, is considered unacceptable because of the scarcity of frequencies and accessible moderate flows. We therefore generally opted for a salvo policy of variable size to sell such VBR flows. While this policy avoids the mess mentioned above, the fact remains that it entails an undesirable delay in transmission.

The object of the invention is to propose a technique for the flow of VBR streams that can both benefit from the advantages of the variable size policy (no bandwidth waste) and the advantages of the fixed size policy (absence additional delay in transmission, facilitation of frequency shift handover).

For this purpose, the invention proposes, according to a first aspect, a method for generating a data stream comprising a plurality of elementary streams temporally cut to be transmitted in the form of bursts, at least one time indicator indicating in a burst of an elementary flow the start time of the next burst of the same elementary stream, the method comprising the steps according to which: - at the latest at the emission of a burst, the starting time of the next burst of the same elementary stream, and the time stamp of the burst is informed as a function of the predicted departure time for the next burst, - at the predicted departure time, the start of transmission of the next burst is forced, even if a salvo another elementary stream is then being transmitted, and in this case then a multiplex transmission of said next burst with said burst of another elementary stream being transmitted. Some preferred, but not limiting, aspects of this method are the following: during the multiplex transmission of said next burst with said burst of another elementary stream being transmitted, x% of the bandwidth allocated to the transmission of the data stream is allocated to said next burst and (100-x)% of said bandwidth is allocated to said burst of another elementary stream being transmitted;

at the predicted start time of said next burst, a packet of said next burst is transmitted and x = 0% is considered;

the multiplex transmission is implemented until the end of the transmission of said burst of another elementary stream; - The multiplex transmission is implemented for a limited time;

if, at the end of the limited time, sections of said burst of another elementary stream have not been transmitted, said sections are reported to be transmitted at the beginning of the next burst of said other elementary stream;

a level of service is guaranteed for each elementary stream, and the start time of a burst of an elementary stream is predicted to correspond to each elementary stream its guaranteed level of service; the guaranteed level of service for a basic stream is a guaranteed level of service for the transmission of each of the bursts of the elementary stream;

- the guaranteed level of service for the transmission of a burst is a guaranteed number of packets for the transmission of burst data; the predicted start time of the next burst corresponds to the sum of the number of packets guaranteed for the transmission of the bursts of each of the elementary flows;

- the guaranteed level of service for an elementary stream corresponds to the average, minimum or maximum level of service of a variable-sized elementary stream;

a burst comprising a plurality of sections, the header of each section comprising a time indicator indicating the starting time of the first section of the next burst of the same elementary stream, for each section the duration is calculated between the date of transmission of the section and the predicted departure time, and information is provided on the temporal indicators of the sections with the said calculated durations;

the data stream is a DVB-H compatible stream. According to a second aspect, the invention relates to a device for processing a data stream comprising a plurality of elementary streams temporally cut to be transmitted in the form of bursts, and comprising means for implementing the method according to the first aspect. of the invention. 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 drawings in which: :

FIG. 1 represents the delta-t temporal indicators contained in the header of each of the MPE sections of a burst of an elementary stream to indicate the departure time of the first section of the next burst of the same elementary stream. ;

FIG. 2a shows a burst transmitted alone on the transmission channel according to the usual method, and FIGS. 2b and 2c show a burst transmitted in accordance with the invention using the possibility of multiplexing during part of the transmission of the burst;

FIG. 3 is a diagram illustrating the transmission of the bursts of different elementary flows in accordance with a possible implementation of the method according to the first aspect of the invention.

The invention is based on a transmission system in which a data stream comprises a plurality of ES elementary stream services or streams (temporally sliced) to be transmitted in the form of bursts. A preferred, but not limiting, example of such a system is a system compatible with the DVB-H standard.

The invention relates in particular, but not exclusively, to the transmission of elementary variable rate flow VBR ("Variable Bit Rate"), in particular video streams. A VBR flow is characterized by its average flow Vav, its maximum flow Vmax, and its minimum flow Vmin.

As will be detailed later, the invention proposes to use the possibility of multiplexing bursts of different elementary streams on the transmission channel.

The DVB-H standard actually allows multiplexing of bursts. As seen previously, a 500kbps average rate service can be sent on a 10Mbps channel with bursts of 100ms duration repeated every two seconds. If we consider two services of the same type, we can send them sequentially, each occupying 10Mbs of the channel for 100ms, or multiplex them at the MPEG-2 or MPE level so that each service will have a duration Bd of 200ms and will occupy 5Mbs on average.

Multiplexing bursts leads to an increase in the duration of the bursts, and consequently a decrease in the electrical autonomy of the terminal. On the other hand, the multiplexing of the bursts makes it possible to increase the quality of the transmission insofar as a loss of information during a given duration will touch data belonging to bursts. different and therefore translate at the level of each burst by fewer bits lost. The error correction mechanisms (MPE-FEC typically) associated with each burst will then have a better chance of recovering the error. FIG. 1 shows the time indicators contained in the header of each of the MPE sections of a burst B, _j of an elementary stream ES, (burst No. j of the service No. i) to indicate the departure time t, _j + i of the first section of the next burst B _{IJ +} i of the same elementary stream ES ₁ . In this figure, the burst B _1J of the elementary stream ES ₁ comprises five sections MPE S1 -S5. A time-delta-t indicator S1 -delta-t S5 is indicated in the header of each section S1 -S5, to indicate therein a time information relating to the departure time t _{IJ +} i of the first section of the next burst B _{IJ +} i of the same elementary stream ES ₁ . This is relative time information in that it informs that the next burst will start x milliseconds from now.

We have seen previously that in a salvo policy of fixed size, these temporal indicators are easily calculable and can be informed without delay during the constitution of the salvo.

On the other hand, in a salvo policy of variable size, these temporal indicators can only be filled in after knowing the effective start time of the next burst, which requires having waited for the instant t _{IJ +} i to obtain a complete view of the bursts between the current salvo and the next salvo.

In the context of the invention, the temporal indicators (delta-t S1 -delta- t S5) of the sections of a salvo are given at the latest during the constitution of the salvo, on the basis of a prediction of time from the next burst of the same elemental stream. Thus, at the latest at the emission of a burst B, _J , the departure time t _{IJ +} i of the next burst B _{IJ +} i of the same elementary stream ES ₁ is predicted, and informs the time indicator (s) of the burst B _1J as a function of the predicted departure time for the next burst B _{IJ +} i.

Then, at the departure time predicted t _{IJ +} i, it forces the start of the transmission of the next burst B _{IJ +} i, including if a burst B _kJ of another elementary stream ESk is then being transmitted.

In particular, if a burst B _kJ of another elementary stream ESk is being transmitted at the start time predicted t _{IJ +} i of the next burst

B _{IJ +} i of the stream ES ₁ , a multiplex transmission of said next burst B _{I J +} i is carried out with said burst B _kJ of another elementary stream (ESk) being transmitted.

And it is repeated, at the latest on the emission of the next burst B _{I J +} i of the stream ES ₁ , the prediction step for evaluating the departure time t _{IJ + 2} of the next burst B _{1J + 2} of the same stream ES ₁ , and inform the time indicator (s) of the salvo B _{IJ +} i. FIG. 2a shows a burst B1 transmitted alone on the transmission channel according to the usual method, this burst being characterized by a rectangle of dimensions Bb (bandwidth), Bd (duration)

FIGS. 2b and 2c show a burst B1 transmitted in accordance with the invention, using the multiplexing possibilities during part of the transmission of the burst. The burst B1 is characterized by a figure of the same surface as that of FIG. 2a

(equality of the number of bits transmitted by the burst).

FIG. 2b illustrates a general case, while FIG. 2c represents a particular case in which the figure characterizing the salvo B1 is composed of a right triangle T1 (whose right-angle forming cathets are Bd and d1), a rectangle R (of dimensions Bb, d2) and a triangle T2 (whose catheters forming the right angle are Bd and d3).

In the example shown here, the transmission of the burst B1 is broken down into three periods. A first period of duration d1, initiated at the start time t0 predicted for the transmission of B1 at most during the transmission of the previous burst of the same stream, during which the burst B1 and a salvo BO of the previous elementary stream are transmitted from multiplexed way on the channel. A second period d2 during which the burst B1 is transmitted alone on the channel.

A third period d3, initiated at the predicted start time t1 for the transmission of a burst B2 of the next elementary stream, during which the burst B1 and the burst B2 of the next elementary stream are transmitted in a multiplexed manner on the channel.

The duration of transmission of the burst B1 goes from Bd (Figure 2a) to Bdv = d1 + d2 + d3 (Figure 2c).

The equality of the surfaces (the number of bits constituting the burst) leads to the following equality: Bd = d2 + (d1 + d3) / 2. From this we deduce BdV = Bd + (d1 + d3) / 2.

This increase in the duration of the salvo will result in a decrease in the battery economy on the terminal. But as we saw earlier, better protection against error can be obtained due to the implementation of multiplexing bursts on the channel. In the context of the invention, the distribution of the departure times t, _j can be chosen arbitrarily, or, as will be presented hereinafter, with a spacing between the starting times of two consecutive bursts (t i + 1). _j - t, _j ) = Ni ^* T translating a level of service associated with the different services. It will be understood that in the context of the invention, it is important that the departure times t _{IJ +} i are predicted from burst n ° j of the same service n ° i. This does not necessarily mean that the spacings (t, ₊ i _j - t _u ) remain fixed in time. The invention also extends to the adaptation of these spacings (t, + i, _j - 1 ,,,) over time, for example as a function of the flow rates of services actually observed. According to one possible embodiment of the invention, each elementary stream is guaranteed a level of service.

The starting times of the successive bursts of the different services are then predicted so as to be distributed so as to guarantee the level of service associated with each elementary stream.

According to a preferred embodiment, it is in particular to guarantee a given level of service for the transmission of each bursts of the same elementary stream.

Guaranteeing a given level of service for the transmission of a burst can for example consist in guaranteeing a number of MPEG packets for the transmission of the burst (or in other words to guarantee a duration for the transmission of the burst, that is to say this corresponds to the time required to transmit the guaranteed number of packets).

The guaranteed level of service for an elementary stream can be set arbitrarily. It may also correspond to the average, minimum or maximum level of service of a VBR elementary stream.

And as we have seen previously, this level of service can be modulated during the transmission, for example to take into account the rates actually recorded for each of the elementary flows.

In particular, it is possible to guarantee a number of packets for the transmission of a burst of an elementary stream, this number being chosen, for each elementary stream, on the basis of:

the average number of packets per burst of the elementary stream, itself derived from the average bit rate (Vav) of the elementary stream VBR, or

the minimum number of packets per burst of the elementary stream, itself derived from the minimum bit rate (Vmin) of the elementary stream VBR, or

the maximum number of packets per burst of the elementary stream, itself derived from the maximum bit rate (Vmax) of the elementary stream VBR. Referring to Figure 3, we consider in the following an example of a data stream to be transmitted comprising three elementary streams ESi, ES2, ES _3.

The transport stream will be of type Bi _{ι 0>} B ₂ , o, B ₃ , o, Bi, i, B ₂ , i, B ₃ , i, Bi _ι ₂ _> B ₂ , 2, B ₃ , 2, --- - B _1J , where B _1J represents the burst n ° j of the elementary flow n ° i.

For example, a guaranteed level of service is associated with an elementary stream by guaranteeing a number of packets for the transmission of bursts of this elementary stream. For example, we can choose to guarantee:

N1 = 40 packets for the transmission of each of the bursts of the elementary stream n ° 1;

N2 = 30 packets for the transmission of each of the bursts of the elementary stream n ° 2;

- N3 = 60 packets for the transmission of each bursts of the elementary stream n ° 3. In the context of the invention, at the latest on issuing a burst B _1J , the start time t _{I J +} i of the next burst B _{I J +} i of the same elementary stream ES ₁ is predicted on the basis of Guaranteed service levels, and the time-delta-t of the burst B _{1J is reported} as a function of the predicted departure time for the next burst B _{IJ +} i. According to one embodiment of the invention, the predicted departure time corresponds to the sum N of the guaranteed packet numbers N1, N2, N3 for transmitting a burst of each of the elementary streams.

The predicted departure times t, _j are thus spaced by a duration reflecting the service levels guaranteed to the various services. We observe in particular the relation (t, + i _{, j} - t, _J ) = Ni * T, where T represents the transmission time of an MPEG-2 packet on the BP _{MAX whole} of the assigned bandwidth BP. to the transmission of the data stream.

In other words, the predicted departure time corresponds to the moment at which the next burst should begin to be issued in the event that in the meantime, each of the bursts was transmitted alone on the channel (or at least only on the part of the channel allocated to the DVB-H) respecting the level of service that has been associated with it.

It is thus predicted that the start time t _{IJ +} i of the burst B _{iJ +} i is located at N ^* T seconds of the departure time t ,,, of the burst B _1J . It is actually considered that the bursts of the three elementary streams to be transmitted before the next burst of the stream ES ₁ will each be successively transmitted using the guaranteed number of packets.

Taking the example of the burst B ₂ , i, it is predicted that its start time t.2,1 is located after transmission of N = N2 + N3 + N1 packets (the N2 packets of the burst B ₂ , o + the N3 packets of burst B ₃₀ + N1 packets of burst Bi, i), ie N ^* T seconds after the beginning of salvo B ₂ , o to t ₂ , where

Bursts consisting of a plurality of sections S1-S5 are considered in the following.

Considering the burst Bi ₀ , the delta-t time indicator S1 of the first section S1 is informed at the emission (at t _S i) of said section S1 according to: delta-t S1 = tι, i-t _S i

In the same way, we inform the time-delta-t indicator Sm (m = 2, ..., 5) of a following section Sm, at the emission (at t _Sm ) of said sections S2-S5, according to delta -t Sm = ti, i - t _Sm . Thus, for each section of a burst, the duration between the date of transmission t _Sm of the section and the predicted departure time for the next burst B _{IJ +} 1 of the same service ES ₁ is calculated, and the delta-T of the section of the burst B _1J with said calculated duration.

Returning to the general description of the invention, force, predicted start time t _{IJ +} i, the start of transmission of the next burst B _{IJ +} i, including whether a burst of another elementary stream is then in progress of transmission.

In particular, if a burst of another elementary stream is being transmitted to the predicted start time of said next burst, a multiplex transmission of said next burst with said burst of another elementary stream being transmitted is carried out. . In other words, if the elementary streams have transmitted, in the interval between the transmission starts t, _j , t _{IJ +} i, two consecutive bursts B _1J , B 1, _{J +} 1 of the same elementary stream ES 1, plus that what was allocated to them (for example an arbitrary service level, or a level of service corresponding to the average service level of a VBR stream), a multiplex transmission is carried out.

There is shown in Figure 3 different multiplex transmissions of bursts of different services.

For example, at the time t ₂ , o of the predicted start of the burst B ₂ , o (burst n ° 0 of the service n ° 2), the burst Bi ₀ (burst n ° 0 of the service n ° 1) is still in progress of transmission. The start of the burst B ₂ , o is then forced at t ₂ , o, and a multiplex transmission is carried out on the part of the channel allocated to the transmission of the data stream (this part having a bandwidth BPMAX) of burst B ₂₀ with burst Bi, ₀ . Of course, when a burst of another service is no longer being transmitted to the predicted departure time, there is no need to perform a multiplex transmission. This case is represented in FIG. 3 for the transmission of the bursts B _2, i and Bi ₂ which are transmitted from the outset to t ₂ , i and t i, ₂ only on the channel, with the complete BP _MAX bandwidth.

The multiplex transmission may for example consist, as represented in FIG. 3, in allocating x% of the bandwidth BP _MAX to the incoming burst whose transmission is forced to its predicted departure time and (100-x)% from the bandwidth BP _MAX to the burst still being transmitted.

A limiting case of this embodiment consists of the predicted departure time t, _j + i of the burst B _{IJ +} i, and in the case of multiplex transmission, to force the departure of said burst B _{IJ +} i by transmitting only 'a package and considering x = 0%. The distribution of the bandwidth BP _MAX between bursts can of course be carried out according to different schemes, for example as this is shown schematically in Figure 2c by providing a linear increase in the bandwidth allocated to the incoming burst.

According to a possible embodiment of the invention, the multiplex transmission is implemented until the end of the transmission of the burst of another elementary stream still being transmitted to the start time of a burst.

According to another possible embodiment, and as is apparent in FIG. 3 for the multiplex transmission of the bursts B ₂ , o and B ₃₀ from t ₃ , ₀ , the multiplex transmission is implemented during a limited time W

At the end of this limited period t | _im , only burst B ₃₀ will be transmitted on the channel. And the sections of the salvo B ₂ , o not yet transmitted at the end of this limited time will be reported to be transmitted at the beginning of the next burst B ₂ , i of the same elementary stream ES ₂ (that is to say from t _2, i).

It is understood that the invention implements an "a priori" calculation of the temporal indicator (s) delta-t during the constitution of the burst. The invention thus has the advantage over a policy of bursts of variable size not to introduce delay in the transmission of bursts. It thus benefits from the advantages of a fixed size policy, such as the facilitation of hand-over by phase-shift, and the absence of any additional delay in the issue.

Furthermore, by using the multiplexing possibilities during part of the transmission of a burst, the invention then makes it possible, with respect to the fixed size policy set to maximum, to make better use of the bandwidth resource.

The invention is of course not limited to the method described above, but extends to an electronic device comprising software and / or hardware means configured to implement this method. This device typically resumes the functions of an IP encapsulator

(IP Encapsulator) receiving IP streams, splitting them temporally, calculating intra-burst FEC protection, encapsulating IP packets and embedding them in the transport stream. In particular, the means for encapsulating the IP packets and embedding them in the transport stream are configured to, at the most at the time of issuing a burst, predict the starting time of the next burst of the same elementary stream, to inform the time indicator (s) of this burst, and to force at the start time predicts the transmission of the next burst, including if a burst of another elementary stream is then being transmitted.

Claims

A method of generating a data stream comprising a plurality of temporally-cut elementary streams (ES) for transmission in the form of bursts, at least one time-delay indicator (delta-t) indicating in a burst of an elementary stream the starting time of the next burst of the same elementary stream, the method comprising the steps according to which: - at the latest at the emission of a burst (B,; _J ), the start time (t, _j ) is predicted + i) of the next burst (B _{IJ +} i) of the same elementary stream (ES ₁ ), and the burst time indicator (B _{I; J} ) is informed as a function of the predicted departure time for the next burst (B _{IJ +} i),

at the predicted departure time (t _{IJ +} i), the start of transmission of the next burst is forced, and if a burst (B _kJ ) of another elementary stream (ESk) is then in the process of being transmitted, a multiplex transmission of said next burst (B _{IJ +} i) with said burst (B _kJ ) of another elementary stream (ESk) being transmitted.

The method according to claim 1, wherein during the multiplex transmission of said next burst (B _{IJ +} i) with said burst (B _kJ ) of another elementary stream being transmitted, x% of the bandwidth assigned to the transmission of the data stream is allocated to said next burst (B _{IJ +} i) and (100-x)% of said bandwidth is allocated to said burst (B _kJ ) of another elementary stream being transmitted.

3. The method of claim 2, wherein the start time (t _{IJ +} i) predicts said next burst (B _{IJ +} i), it transmits a packet to said next burst (B _{IJ +} i) and we consider x = 0% .

4. Method according to one of claims 1 to 3, wherein the multiplex transmission is implemented until the end of the transmission of said burst of another elementary stream.

5. Method according to one of claims 1 to 3, wherein the multiplex transmission is implemented for a limited time.

The method according to claim 5, wherein if at the end of the limited time sections of said burst (B _kJ ) of another elementary stream (ESk) have not been transmitted, said sections are reported for to be transmitted at the beginning of the next burst (B _k , _{J +} i) of said other elementary stream (ES _k ).

7. Method according to one of the preceding claims, wherein a level of service (N1, N2, N3) is guaranteed for each elementary stream.

(ESi, ES2, ES3), and the start time (t _{IJ +} i) of a burst of an elementary stream is predicted to correspond to each elementary stream its guaranteed level of service (N1, N2, N3).

The method of claim 7, wherein the guaranteed level of service for an elementary stream is a guaranteed level of service for transmitting each of the bursts of the elementary stream.

The method of claim 8, wherein the guaranteed level of service for the transmission of a burst is a guaranteed number of packets for the transmission of burst data.

10. The method of claim 9, wherein the predicted start time of the next burst is the sum (N) of the number of packets guaranteed for the transmission of bursts of each of the elementary streams.

11. Method according to one of claims 7 to 10, wherein the service level guaranteed for an elementary stream corresponds to the average service level, minimum or maximum of a variable size elementary stream.

The method according to one of the preceding claims, wherein a burst comprises a plurality of sections (S1-S5), the header of each section comprising a time indicator (delta-t S1-delta-t S5) indicating the starting time of the first section of the next burst of the same elementary stream, in which for each section the duration between the date of transmission of the section and the predicted departure time is calculated, and the temporal indicators (delta- t S1 ... delta-t S5) sections with said calculated durations.

The method of any one of claims 1 to 9, wherein the data stream is a DVB-H compatible stream.

14. Device for processing a data stream comprising a plurality of elementary streams temporally cut to be transmitted in the form of bursts, characterized in that it comprises means for implementing the method according to any one of the claims. 1 to 13.