FR2726722A1 - Digital packet routing method e.g. for digital television system - Google Patents

Abstract

The method involves transmitting input data received in a data stream between a system transmission unit and a receiver, with the packets arranged according to the order in which they are placed in the input stream. The packets have been previously numbered in order. Transmission is on different channels according to the level of priority of the input packets. An output packet stream is formed and uses a transmission protocol for a full packet in the output stream. The protocol is incremented each time a numbering error is detected in one of the packets which is ready to be transmitted. The process continues until all the data has been examined.

Description

 The present invention relates to a packet diverting method intended to be implemented in a reception unit of a digital data transmission system arranged in packets.

The invention applies in particular in the case where said digital data are representative of image signals and of sound signals. It therefore belongs, in this application, to the field of digital television.

 A television system is provided for transmitting image signals, sound signals and data signals to receivers which reconstruct the original images and sounds and which use the data appropriately to present these images and sounds . Several programs must be transmitted, each program can include several sound channels.

 In a digital television system, image data and sound data from multiple image and sound sources are encoded to form what is known in the art as elementary image streams and elementary sound streams. These streams are packaged to produce packets that are called PES packets, such as Packetised Elementary Stream, of varying size. The PES packets are divided into transport packets of fixed size which are time multiplexed in order to form a single data stream, called transport stream, for example of the type defined in the standard known as MPEG2.

 It will be noted that each elementary stream corresponds to a sequence of images or sounds which has been coded. A transport stream contains several television programs each consisting of one or more elementary streams of images and one or more elementary streams of sound.

 The transport stream from the multiplexer is then transmitted in asynchronous mode to a packet transmission system consisting of a transmission unit and at least one reception unit.

 In the transmission unit and before its transmission, for example over the air, to the reception units, the transport stream is subjected to a processing which essentially consists of coding, for example a coding for the detection and correction of errors, such as Reed Solomon coding. The coded transport stream is then modulated for transmission.

 In a reception unit of the transmission system, the reviewed signals are demodulated and the transport stream is recovered, the decoding operation corresponding to the coding and correction operation having been carried out beforehand. The recovered transport stream is demultiplexed and each channel carrying an elementary stream is decoded, in a decoder, for the presentation in a presentation unit of the elementary streams of a program.

 In a transport flow, each transport packet has a header, which essentially contains a so-called PID (Packet IDentifier) used to identify the elementary flow carried by this transport packet. Thus, transport packets having the same PID identifier transport the data of a single elementary stream. There are therefore at least as many identifiers as elementary flows present in the transport flow.

 A transport packet is assigned a PID when multiplexing the corresponding PES packet (s). It is used to reconstruct the elementary flows from the transport flow.

 It is interesting and advantageous to separate the useful data of the transport stream into several fields which correspond to different priority levels, this in particular so that the data having priority levels can be treated differently at the time of coding in the unit of transmission, during transmission and at the time of decoding in the reception unit. It results from this assignment of priority levels and from the concomitant coding that the packets of the incoming stream are transmitted between the transmission unit and a reception unit of the transmission system on different transmission channels which are each specific to the levels of priority to which said packets of the incoming flow are assigned.

 These priority levels can for example be defined in the identifiers of the packets of the transport stream. The data of a transport stream can be separated into three fields: high priority data corresponding to the best protected information, such as data relating to standard definition images, medium priority data, as additional data necessary for high definition of the same images, and low priority data corresponding to additional data necessary for very high definition of the same images.

 The use of several priority levels means that the different packets of the incoming flow have to be routed in one or other of the transmission channels which are each assigned to a priority level. This routing operation is carried out in the transmission unit and this before the coding operation mentioned above in order to be able to carry out a specific coding for each transmission channel.

 In the receiving unit, it is necessary to recombine the packets into a single outgoing flow, which must be identical to the incoming flow. This object is notably achieved when the packets of the outgoing flow are found in the same order as that which they occupied in the incoming flow. To achieve this goal, it has been proposed to number the packets of the incoming stream before they are routed to the different transmission channels. This numbering can, for example, consist of the introduction into the data of each packet of a field representative of the serial number of arrival of said packet in the incoming stream. In the receiving unit, a rerouting device is provided which transmits the packets from the transmission channels in the outgoing stream in the order of the numbers to which said packets are assigned.

 This numbering system and the rerouting method which is implemented in said rerouting device make it possible to recover the packets in the outgoing stream in the same order as that which they occupied in the incoming stream. They are entirely satisfactory as long as there are no transmission errors in a number of a packet. Indeed, in this case, the serial number of the packet in question, as the rerouting device can read it, is wrong and the transmission of packets on the outgoing stream is disturbed, or even blocked.

 The present invention aims to solve this problem and, to do this, provides a method of diverting packets present in each of the transmission channels which allows, in the presence of a transmission error in the serial numbers to which the packets are affected, a transmission of the packets on the outgoing flow little disturbed and such that the order of the packets in the outgoing flow is respected relative to that in the incoming flow.

 To do this, a packet diversion method according to the invention is remarkable in that it consists in using a right to send a stuffing packet in the outgoing stream, said right being incremented at each error detection. numbering in one of the packets which are present in said channels and which are ready to be transmitted in said outgoing stream and being decremented during the processing of said numbering error so that said outgoing stream respects the order of the packets said incoming flow.

 According to another characteristic of the invention, it consists in transmitting, at the same time as a numbering error is detected, a stuffing packet in said outgoing stream.

 According to another characteristic of the invention, it consists in deleting packets ready for transmission which are liable to be erroneous, the right to transmission then being incremented by a number of units equal to the number of packets deleted less a.

 According to another characteristic of the invention, said operation of processing said numbering error consists in sending a stuffing packet in said outgoing stream.

According to a variant, a method according to the invention is remarkable in that it consists in searching, among the packets which are present in said channels and which are ready to be transmitted in said outgoing stream, that which carries the number following the number of the package that was just issued, then,
a) if said number has been found only once, to issue said packet,
b) if said number has been found several times, to send a padding packet and, at the same time, to delete the packets likely to be erroneous and to increment by a number of units equal to the number of deleted packets minus one counter whose value
A represents a right to send A jam packets, and
c) if said number has not been found, to send a padding packet and, if said counter does not have a zero value, to decrement said counter by one, while if said counter has a zero value, deleting all the packets ready to be sent and incrementing said counter by a number of units equal to the number of dropped packets minus one.

 In order to minimize disruption of the outgoing flow and avoid deleting packets which do not contain incorrect numbers, it is provided, in step b), that only packets bearing the number following the packet number which has just be read be deleted.

 According to a variant, all the packets ready to be sent are deleted.

The characteristics of the invention mentioned above, as well as others, will appear more clearly on reading the following description of an exemplary embodiment, said description being made in relation to the accompanying drawings, among which:
Fig. 1 is a block diagram of a television system which is equipped with a reception unit in which the method of the invention is implemented,
Fig. 2 is a block diagram of a routing device of a transmission system such as that which is represented in the
Fig. 1,
Fig. 3 is a block diagram of a rerouting device / decoder in which the rerouting method according to the invention is placed,
Fig. 4 is a diagram showing the operation of a transmission system such as that shown in FIG. 1,
Fig. 5 is a flowchart illustrating a rerouting method according to the invention in the case where the number of transmission channels of the transmission system is three,
Fig. 6 is a diagram showing, in eight successive steps, the state of the number to be read, the state of the counter entitled to the transmission of a stuffing packet, the state of the FIFO queues of the diverting device, and
Fig. 7 is a flow diagram illustrating the rerouting method according to the invention in the case where the number of transmission channels of the transmission system is n.

 The television system shown in FIG. 1 essentially comprises an image coder CI, a sound coder CS and a data coder CD, a time multiplexer MUX, a transmission system ST consisting of a transmission unit UEM connected, by a transmission link lt, at least one reception unit URE, a time demultiplexer DEM, a picture decoder DI, a sound decoder DS and a data decoder DD.

The coders CI, CS, CD are designed to deliver, from the elementary streams respectively delivered by a plurality of image, sound and data sources (not shown), data which are transmitted in the form of PES packets to the multiplexer
MUX. The MUX multiplexer is intended for its part to split the PES packets into fixed size packets, called transport packets, and to temporally multiplex said transport packets in order to form a transport stream Fep.

This transport stream Fep is transmitted to the input of the transmission system ST and, more particularly, to the input of the emission unit UEM. I1 is called thereafter incoming packet flow
Fep.

 The incoming stream of Fep packets is transmitted by the transmission system ST and is recovered and delivered at its output in the form of an outgoing stream of Fsp packets.

 The outgoing flow Fsp generated by the reception unit URE is transmitted to the input of the demultiplexer DEM whose outputs are respectively connected to the inputs of the decoders DI, DS and DD. These are intended to decode the data of the elementary streams carried by the packets delivered by the demultiplexer DEM and to present this data in the form of sequences of images and sounds.

As already mentioned above, each packet delivered by the MUX multiplexer and therefore ending up in the incoming flow
Fep belongs, according to the priority level to which the data it contains is assigned, to three types of packets: a high priority type HP, a medium priority type MP and a low priority type LP.

 It is in the CI, CS and CD encoders that the data is assigned a priority level. In the Fep incoming flow, each packet has a header which itself contains the identifier called PID used to identify the elementary flow which it carries.

Thus, each PID identifier is associated with a priority level, HP, MP or LP.

 The UEM transmission unit essentially consists of a ROUT routing device, a first coder COD1, a second coder COD2 and a transmitter EM.

The main function of the ROUT routing device is to separate the incoming flow of Fep packets into as many packet flows
FHp, Fp and FLP that there are priority levels, here three. Another function of the ROUT routing device is to solve the problems resulting from the difference between the bit rate generated by the MUX multiplexer and the bit rate of the ST transmission system, which is necessarily higher, as already mentioned, than the bit rate generated by the MUX multiplexer.

 The first encoder COD1 is for example an encoder of the encoder / corrector type such as a Reed Solomon encoder followed by the interleaving circuits provided for each transport stream of different priority.

 The second coder COD2 is for example a convolutional coder.

 The reception unit URE essentially consists of a receiver REC, a decoder DEC, for example of the convolutional coder type corresponding to the convolutional coder COD2 of the transmission unit UEM, and a decoder / rerouting device DEROUT.

 The function of the DEROUT decoder / rerouting device is, on the one hand, to carry out a decoding which corresponds to the coding carried out by the coder COD1 of the transmitting unit UEM and, on the other hand, to group in a single transport stream Fsp the streams received from packets FrHp, FrZp and FrLp of different priority from the decoder DEC.

 Each transport packet HP, MP, LP in the incoming stream Fep as well as in the streams FHp, Fp and FLP is, in one embodiment of the invention, made up of 188 bytes of which the first 4 constitute its header. In these 4 bytes, 13 bits constitute the PID identifier of the packet used to identify the elementary stream conveyed by this packet.

It will be noted that, the higher the priority level of the data of a transport packet, the more their transmission must be done with a low error rate and the more the loss of their content must be avoided. Also, each separate packet stream FHPT FMp and FLP with a determined priority level is therefore treated separately, in particular in the convolutional coder COD2, in the transmitter EM and the receiver
REC and in the DEC decoder.

 There is shown in FIG. 2 the ROUT routing device according to the invention.

 It includes an EFP input to which the incoming stream of Fep packets from the MUX multiplexer is supplied.

 The EFP input is connected to the input of a FIFOE input queue, the output of which is itself connected to the input of a device for inserting DIPB stuffing packets. The DIPB device is designed to insert stuffing packets in order to adjust the bit rate in the ST transmission system to that of the incoming stream Fep.

The output of the DIPB device is connected to the input of a re-stamping device DR, a control input of which is connected to a stamp covering circuit CR. The latter has its input which is connected to the input EFp of the routing device
ROUT.

As is known in the art, to synchronize the decoding and presentation operations of the DI, DS, DD decoders (see Fig. 1), time stamps indicating the correct times for presenting and decoding the images and the its partner as well as stamps indicating the values of a reference clock are incorporated in the packets
PES and therefore in the packets of the incoming flow F. The first are called PTS and DTS respectively (as Presentation
Time Stamps and Decoding Time Stamps). The seconds are called PCR (like Program Clock Reference).

 The PCR stamps indicate the correct values of the system reference clock and allow the reference clock signal to be recovered in the DI, DS and DD decoders.

 The DTS decoding stamps and the PTS presentation stamps indicate the instants which are defined with respect to the system reference clock and at which the decoding and presentation processes of the elementary streams recovered in said decoders must be triggered.

 It will be understood that the insertion of the padding packets by the DIPB device leads to recalculating the values of the PCR stamps.

To do this, the circuit CR receives the HP, MP and LP packets from the incoming packet stream Fep extracts the time stamps
PCR and retransmits them to the DR re-stamping device which calculates new stamp values and which assigns them to predetermined packets originating from the stuffing packet insertion device
DIPB.

 The output of the DIPB device is connected to the input of a NUM packet numbering device whose essential function is to assign to each of the packets on its input a number which corresponds to that of its order in the packet flow to the output of the DIPB tamper packet inserter. These numbers are, for example, entered in a field provided for this purpose in each of the packages.

 The output of this NUM device is connected to the input of a DAP packet routing device, the three outputs of which are respectively connected to the three inputs of three FIFON, FIFOM and FIFOL queues. The outputs of these three FIFOH, FIFON and FIFOL queues are linked to the SR output of the ROUT routing device.

 It will be noted that the separate packet streams FHp, Fp and FLp are, at the output of the routing device ROUT, carried by the same time multiplex noted in FIG. 2 (FHPT Fp, FLp).

 It will also be noted that the output SR of the routing device ROUT is intended to be connected to the coder COD1.

 A control unit UC has an output connected to the control input of the packet routing device DAP to deliver there a control signal Cdap. It also has six outputs respectively connected to the write and read command inputs of the three FIFOH, FIFoh and FIFOB queues to respectively deliver write command signals WH, WM and We and read command signals RH, Rz and RB. It also has two outputs respectively connected to the write command input and to the write command output of the FIFOE queue to respectively deliver the signals Sce and Scl to them.

 The control unit UC also has two inputs respectively designed to receive, on the one hand, the clock signals Hm delivered by the multiplexer MUX in relation to the system clock of the encoders CI, CS, CD and in synchronism with the packets of the incoming stream Fep and, on the other hand, of the system clock signals Hs. Finally, it includes an input on which signals of operating modes Mf are delivered.

 The DEROUT decoder / rerouting device associated with the ROUT routing device is now described in relation to FIG. 3.

I1 comprises three decoders DRSH, DRSM and DRSL corresponding to the coder COD1 of the transmitting unit UEM whose inputs are respectively connected to the corresponding outputs of the convolutional decoder DEC to respectively receive there the streams of packets FrHp, Frp and FrLp carrying respectively high priority packets HP, medium priority packets
MP and low priority packets BP. The outputs of the three decoders are connected to the input of three queues FIFOrH, FIFOrM and FIFOrL, the outputs of which are connected to the input of a DM rerouting device, which has its output which constitutes the SFP output of the decoder / DEROUT rerouting device on which the outgoing flow of Fsp packets is delivered.

 The DM rerouting device has a control input which is connected to a control output of a control unit UCr to receive a signal Cm there. The control unit UCr is also provided for delivering read signals Rh, Rm and Rl to the queues FIFOrH, FIFOrn and FIFOrL. It is also designed to receive the system clock signal Hs as well as a control signal for operating modes Mf.

We will now describe the operation of the ROUT routing device and the DEROUT rerouting device in relation to the
Fig. 4. But beforehand, we will specify the characteristics of the frames transmitted in the ST transmission system.

 The transmission unit UEM and the reception unit URE operate so that, for each priority level, an integer number of transport packets HP, MP, LP is transported per frame. Several operating modes can be envisaged, corresponding to two bit rates of 12 and 24 Mbits / second respectively and to the hierarchical aspect or not of the transmission system. For each of the modes considered, the number of high priority HP, medium priority MP and low priority BP low priority BP packets per frame transmitted is listed in the following table.

HP MP LP mode
Mode lA 123 189 211
Mode 1A 123 189 211
Mode 1B 124 372 0
Mode 2A 124 124 0
Mode 2B 124 124 0
Mode 3A 492 0 0
Mode 3B 496 0 0
Table 1
Table 1
It is noted that the modes 3A and 3B are non-hierarchical modes unlike the other modes. We also note that mode 1A is the only one which operates three priority levels.

 These different modes are given by way of example. In particular, in the context of the invention, the number of priority levels is not limited to three but can be any, greater than one.

The control unit UC of the transmitting unit UEM and that UCr of the receiving unit URE are designed to be configured in one of the modes defined above by positioning appropriate signals
Mf on their configuration command inputs.

In Fig. 4, the transport packets HP, MP and LP have been shown in the incoming stream Fep at the output of the time multiplexer MUX, in the queues FIFOH, FIFON and FIFOL of the routing device ROUT after their numbering in the numbering device DN, in the multiplex (FHp, Fw, FLp) at the output of said FIFOH, FIFOM and
FIFOL, then, after reception by the receiver REC and decoding in the decoder DEC, in the FIFOrH, FIFOrn and FIFOrL queues of the decoder / rerouting device DEROUT.

 The present invention is concerned with the reading of the FIFOrH, FIFOrn and FIFOL queues. As will be seen below, this reading is done by commanding the FIFOrH, FIFOrM and FIFOrL queues in the order of the numbers which have been assigned to the packets by the NUM numbering device of the transmission unit UEM. There is shown in FIG. 4, the packets in the outgoing stream Fsp after reading the FIFOrH, FIFOrn and FIFOL queues.

The problem that arises is that errors can occur in packets transmitted by the transmission system
ST. However, these errors can affect the numbers of the packets registered in the packets. I1 then follows, without any measure to remedy it, a blocking of the reading of the FIFOrH, FIFOrM and FIFOrL queues since the order of reading of the packets is disturbed.

 To solve this particular problem, the invention proposes an operation of the control unit UCr which is illustrated by the diagram in FIG. 5.

It is assumed, at the start, that is to say at point A, that the reading of the packet of order N has been carried out correctly. In 10, the sequence number N is incremented by one, then in 20, the new number
N is sought in the three packets which can be read in the FIFOrH, FIFOrn and FIFOrL queues.

 If it exists, three cases can arise depending on whether the number of readings of the number sought in 30, is one, two or three.

 It can come in one of three packages. This is the normal case. The corresponding package is read, at 40, and the process starts again at point A.

 It can come in two of the three packages. In this case, there is an error in the numbering of either package. The packets in question are then not read and are eliminated at 50. At the same time, a stuffing packet is sent, at 60, and a counter C is incremented by one, at 70. The process starts again at A .

 Finally, it can come in all three packages. In this case, there is an error in the numbering of two of the three packets. The three packets are not read and are discarded at 80. At the same time, a stuffing packet is sent, at 90, and a C counter is incremented by two units, at 100. The process starts again at A.

 In 20, the new order number N = N + 1 to be read may not exist in any of the three packets likely to be read in the FIFOrH, FIFOr and FIFOrL queues. In this case, there is necessarily an error, in at least one of the three packets.

 Two cases can then arise depending on whether the counter C is zero or not.

 If it is zero, then the three packets are eliminated at 110, a stuffing packet is sent at 120 and the counter C is incremented by two units at 130. The process starts again at A.

 On the other hand, if it is not zero, a stuffing packet is sent at 140 and the counter is decremented by one unit at 150. The process starts again at A.

 We will illustrate, by an example and in relation to FIG. 6, this process.

 Fig. 6 shows 8 states of the rerouting device in the case where two errors have occurred in the three order packet MP03 which has become MP04 and in the ten order packet HP10 which has become HP08.

 In state I, the HPO1 and HP02 packets have been read. N is then equal to 3. Following the error in the packet MP03 which is actually marked MP04, there is no packet assigned with a number 3 to be read. The result of the process according to the invention is given in state II. The three packets HPO9, MP04 and LP06 are eliminated, the counter C is incremented by two units and a stuffing packet BP is sent.

In state III, it can be seen that the reading of the packets MP04 and MP05 then continued normally. The packet to be read is then the one marked with the number 6 but which is absent since it has been eliminated in state II. A BP jam packet is then sent but the counter C is decremented by one unit (state IV)
The LP07 packet is read normally, then N becomes equal to 8. There, two packets with the number 8 are present. The two packets assigned the number 8 are discarded, a packing packet BP is sent and the counter C is incremented by one unit (State VI).

 In state VI, N is equal to 9. A padding packet BP is sent, the counter C is decremented.

 In state VII, N is equal to 10. Again, a padding packet BP is sent and the counter C is decremented by one.

 It will be noted that in state VIII, the packets stored in the FIFOrH, FIFOrn and FIFOrL queues are not affected by an error, that it does not miss any and that everything is back to normal. It will also be noted that the counter C has again become equal to O. It turns out that the counter C permanently indicates a right to send a stuffing packet.

 Finally, it will be noted that the stuffing packets are inserted into the frame exactly in place of the packets which have been eliminated so that the transport stream delivered to the demultiplexer 50 is still valid.

 In Fig. 7, there is shown a flowchart which illustrates the method of the invention in the case where the number of priority levels is equal to n. In this case, it will be understood that the number of transmission channels necessary for the transmission of the packets each assigned to a priority level is also equal to n. This is the case with the number of FIFO queues of the sending unit UEM and the number of FIFOr queues of the receiving unit URE.

 The diagram in FIG. 7 differs from that of FIG. 5 in that the number N sought, if it exists, can be found 1 time, 2 times, 3 times, ..., n times. At each branch for which it is found a number of times greater than 1, there is transmission of a stuffing packet in 50, 80, ..., 160, there is elimination of the packets which are in the FIFOr queues in reading position and which have the same number, and this in 60, 90, ..., 170 and there is incrementation of the counter C by a number of units equal to the number of times the said number is met minus one , respectively in 70, 100 and 180.

Claims (7)

 1) Packet diversion method intended to be implemented in a reception unit (URE) of a transmission system (ST) of digital data arranged in packets, said packets previously numbered according to the order they occupy in an incoming flow (Fep) then being transmitted between the transmitting unit (UEM) of said system (ST) and a receiving unit (URE) on different transmission channels specific to the priority levels to which said packets of said flow are assigned incoming (Fep) said method consisting in forming an outgoing packet stream (Fsp) from packets transmitted on said channels, characterized in that it consists in using a right to send a stuffing packet in the stream outgoing, said right being incremented at each detection of a dialing error in one of the packets which are present in said channels and which are ready to be transmitted in said outgoing flow and being decremented tee during the processing of said dialing error so that said outgoing flow respects the order of the packets of said incoming flow.  2) Method according to claim 1, characterized in that it consists in transmitting, at the same time that a numbering error is detected, a jam packet in said outgoing stream.  3) Method according to claim 2, characterized in that it consists in deleting the packets ready to be transmitted which are likely to be erroneous, the right to the emission then being incremented by a number of units equal to the number of deleted packages minus one.  4) Method according to claim 1, 2 or 3, characterized in that said operation of processing said numbering error consists in sending a stuffing packet in said outgoing stream.  5) Packet diversion method intended to be implemented in a reception unit (URE) of a transmission system (ST) of digital data arranged in packets, said packets previously numbered according to the order they occupy in an incoming flow (Fep) then being transmitted between the sending unit (UEM) of said system (ST) and a receiving unit (URE) on different transmission channels specific to the priority levels to which said packets of said incoming flow are assigned ( Fep) said method consisting in forming an outgoing packet stream (Fsp) from the packets transmitted on said channels, characterized in that it consists in searching, among the packets which are present in said channels and which are ready to be transmitted in said outgoing stream (Fsp) the one which carries the number following the number of the packet which has just been sent, then,  a) if said number has been found only once, to issue said packet,  b) if said number has been found several times, to send a padding packet and, at the same time, to delete the packets likely to be erroneous and to increment by a number of units equal to the number of deleted packets minus a counter whose value A represents a right to the emission of A jam packets, and,  c) if said number has not been found, to send a padding packet and, if said counter does not have a zero value, to decrement said counter by one, while if said counter has a zero value, deleting all the packets ready to be sent and incrementing said counter by a number of units equal to the number of dropped packets minus one.  6) Method according to claim 5, characterized in that in step b), only the packets bearing said number are deleted.  7) Method according to claim 5, characterized in that in step b), all the packets ready to be transmitted are deleted.