FR2781630A1 - Packet resequencing and spacing method for radio link, involves assigning rank in sequence of transmitted packets belonging to same connection, and scheduling at receiver based on time for output in multiplex - Google Patents

Abstract

Packets belonging to the same connection are assigned a rank on transmission and are attributed a scheduling time at the receiver corresponding to their intended output time in a multiplex. The scheduling time is based on that of packets of the same connection having a lower rank, in such a way that at least a predetermined delay spacing is maintained between them. Preferably the fixed or variable rate of the data is taken into account in attributing scheduling time.

Description

The present invention relates to a resequencing method and

  spacing packets from at least one connection transmitted by a transmitter and received by a receiver in view of their delivery on an output multiplex. Such a method is intended to be implemented in a communication system, in particular a wireless communication system. A communication system generally consists of one or more receivers and a transmitter connected to said or each of said receivers by means of transmission links, for example, if said system is a wireless communication system, by means of radio links. The system to which the present invention applies is designed so that the data which is transferred from a transmitter to said receiver is

  come in the form of packages.

  Generally, communication systems, a fortiori wireless communication systems, are subject to disturbances and interference so that it is necessary to provide for the implementation of an automatic repetition mechanism in the event of loss of power. 'a packet, a mechanism known as ARQ (Automatic Repeat

  Request meaning Request for an automatic repeat).

  An example of implementation of this mechanism is illustrated in FIG. 1. The left vertical line E in this Fig. 1 is a diagram showing, on the transmitter side, the transmission of packets as a function of time (time flow down) and that on the right R is a diagram showing, on the receiver side, the reception of packets transmitted also as a function of time. As for the central part L between the two lines

  vertical E and R, it represents the transmission of packets on the transmission link.

  The packets of a connection are sent by the transmitter in sequence, with sequence order numbers or ranks assigned to them respectively. In this Fig. 1, it can be seen that the packet Io, originating from a virtual connection VC and transmitted on the transmission link L, has been received correctly, on the receiver side. It is therefore delivered on a multiplex S at the output of the receiver. In addition, a positive Acko acknowledgment is

  transmitted to the transmitter. The package was not received correctly and no charge

  reception is therefore not transmitted. As for the I2 packet, it was received correctly, but not being in the order of the sequence, it is retained. Ack2 acknowledgment is

transmitted to the transmitter.

  At the end of a time t, maximum waiting for the acknowledgment after the first transmission of the packet II, the transmitter re-transmits the packet II. Then, in the case of FIG. 1, it is received correctly and is then delivered to the output multiplex S. The

2 2781630

  I2 packet which had been received correctly but which had been retained is also

now issued.

  It can be seen that the packets Io to I2 are delivered on the output multiplex S in

  the order in which they were issued by the issuer although they were received in a different order.

  The operation which consisted in retaining the I2 packet until the correct reception of the packet

It is called resequencing.

  The mechanism which has just been described is generally called "automatic repetition process with positive acknowledgment". This process is of the

  selective repetition (Selective Repeat in English terminology).

  We know of another process of the same type which is acknowledged to it with negative reception. According to this process, when a packet is correctly received, on the receiver side, but it is not in the order of the sequence, it is that at least one packet was not received correctly. For these lost packets, a negative acknowledgment is sent to the transmitter which then triggers the retransmission. Packets received correctly but with sequence numbers out of the sequence are retained and are the subject,

  as before, of a resequencing.

  Note that these automatic repetition mechanisms can also be implemented when the packets to be transmitted belong to several connections

different virtual ones.

  As we can see, it results from these automatic repetition mechanisms and, in particular from the resequencing operations which they imply, more or less significant jitter phenomena which can be the cause of creation.

  bursts of packets, called "burst" in the technical field.

  It is known to use devices which, subjected to a stream of input packets, packets which may belong to different virtual connections, deliver the same packets so that, on the one hand, a minimum time interval T between two consecutive packets of the same connection and, on the other hand, a minimum time interval t between consecutive packets not necessarily belonging to the same connection. These devices are generally called spacers or "schedulers"

in the technical field.

  It will be noted that the automatic repetition mechanism is carried out by only taking account of the sequence numbers of the packets in the sequences which are transmitted by the transmitter whereas the spacing mechanism is carried out by taking into account only

  ranks of packets among packets that belong to the same connections.

3,2781630

  The aim of the present invention is to propose a method which makes it possible to carry out both the resequencing function, when the latter is necessary, and the

  spacing function generally implemented after receiving packets.

  To this end, a resequencing and spacing method according to the invention is of the type o, to each packet, a rank is assigned in the sequence of packets which are transmitted by said transmitter and which belong to the same connection. It is characterized in that it consists in allocating to each packet received by said receiver a programming time corresponding to the time at which it will be delivered on said output multiplex and determined as a function of the programming times allocated to packets of the same connection of ranks lower than hers and so that said times of

  programming are spaced from each other by at least a predetermined duration.

  According to another characteristic of the invention, it consists, if the traffic on the connection to which belongs said packet just received by said receiver is at constant speed, to allocate a programming time equal to the programming time allocated to the first packet of the sequence of packets from the same connection added to a number of inter-packet time intervals equal to the difference of the ranks of said just received packet and of the first packet of the packet sequence. The programming time of the first packet in the sequence of packets which are transmitted by said transmitter and which belong to the same connection is advantageously equal to its

  reception time by said receiver plus the maximum waiting time.

  On the other hand, if the traffic on the connection to which said just received packet belongs is of variable speed, it consists - if the packet of the same connection of rank just lower than said just received packet has already been received, to assign to said just received packet a programming time equal to the programming time already determined for said packet of just lower rank added to a time equal to an inter-packet time interval, unless the date of reception of said just received packet is later than this programming time and , in this case, to allocate to said just received packet a programming time equal to this reception time, and - if the packet of the same connection of just rank lower than said just received packet has not yet been received, to allocate to said packet just received a programming time equal to the time of reception of said packet just received added to a

predetermined time.

4,2781630

  According to a variant, if the traffic on the connection to which said just received packet belongs is of variable speed, it consists - if the packet of the same connection of rank just lower than said just received packet has already been received, to assign to said packet a programming time equal to the longest time between, on the one hand, the programming time already allocated to said packet and, on the other hand, the reception time of said packet, - if the packet from the same connection of rank just lower said just received packet has not yet been received, to assign to said just received packet a programming time equal to its reception time added to a predetermined delay, then to assign to each of the packets of the same connection of rank higher than rank said just received packet a fictitious programming time which is equal to the programming time allocated to said just received packet to which is added a number of equal inter-packet time intervals the number of packets between the packet just received

and the package concerned.

  According to another variant, if the traffic on the connection to which said just received packet is of variable speed, it consists - if the packet of the same connection of rank just below said just received packet has already been received, to assign to said packet a programming time equal to the longest time between, on the one hand, the programming time already allocated to said packet and, on the other hand, the reception time of said packet, - if the packet from the same connection of right rank less than said just received packet has not yet been received, to allocate to said just received packet a programming time equal to its reception time added to a predetermined delay, then to assign to each packet a number of packets of higher rank the rank of said just received packet equal to the number of packets likely to be received during the maximum waiting time, a programming time equal to the programming time of said just received packet auq a number of inter-packet time intervals equal to the number of packets between the packet just received and the packet concerned is added. For example and advantageously, it consists in assigning to the first packet of the sequence of packets which are transmitted by said transmitter and which belong to the same connection a programming time which is equal to its time of reception by said receiver.

2781630

  Advantageously, said delay corresponds to the maximum time during which the packet can be retransmitted at the level of the transmitter. In addition, said delay may be zero in the event that the packet from the same connection as said just received packet but of just lower rank will no longer be taken into consideration if it were received. In addition, the said period is reduced by an amount depending on the number of retransmissions of which the said subject was

  package just received before receiving it.

  According to another characteristic of the invention, it consists in assigning to all the packets of the same connection of rank lower than said just received packet and to which a definitive programming time has not been allocated, a fictitious programming time which is equal the programming time of said just received packet from which is subtracted a number of inter-packet time intervals equal to the number of packets between the just received packet and the concerned packet and, when the date of receipt of a just received packet is later than said time fictitious programming already assigned to said package, to

destroy said package just received.

  According to another characteristic of the invention, it consists in assigning to packets of the same connection of lower rank to said just received packet, which have not yet been received and which can still be received and taken into account a fictitious programming time. which is equal to the programming time of said just received packet from which is subtracted a number of inter-packet time intervals equal to the number of packets between the just received packet and the concerned packet and, when the date of receipt of a just received packet is after said fictitious programming time already allocated

  said package, to destroy said package just received.

  According to another characteristic of the invention, it consists, if the rank of the packet just received is greater than the highest rank of packets of the same connection already received by a number corresponding to the number of packets likely to be received during the maximum waiting time, to be allocated to said just received packet a programming time equal to the longest time between, on the one hand, the reception time added to a predetermined time corresponding to the maximum waiting time and, on the other hand share, the programming time allocated to the packet with the highest rank of packets of the same connection received so far plus a number of inter-packet time intervals equal to the number of packets likely to be received

  during the maximum waiting time plus one unit.

  According to another characteristic of the invention, it consists in storing each packet rank number received by said receiver in the box of an associated memory

6 2781630

  to the connection to which said packet belongs, said box being associated with the rank of said packet in the sequence of packets which are transmitted by said transmitter and which belong to the same connection, to memorize in said box the programming time allocated to said packet, then delivering said package to a spacer. It further consists in associating with each box of said memory a flag indicating whether the programming time associated with said box can still be changed or not. It also consists in positioning a signaling bit in each

  package which indicates whether or not said package will be taken into consideration.

  The features of the invention mentioned above, as well as others,

  will appear more clearly on reading the following description of an example of

  realization, said description being made in relation to the attached drawings, among

  which: Fig. 1 is a diagram illustrating the transmission and reception of packets called, in the prior art, with automatic repetition ARQ, FIG. 2 is a diagram illustrating the transmission and reception of packets in a system implementing the resequencing and spacing method according to the invention, FIG. 3 is a diagram illustrating an example of the transmission and reception of three packets with implementation of the resequencing and spacing method according to the invention, FIG. 4 is a diagram illustrating a first embodiment of a method according to the invention, FIG. 5 is a diagram illustrating a second embodiment of a method according to the invention, FIG. 6 illustrates the reception of 8 packets as a function of time serving as an example illustrating the implementation of a method according to the invention, FIG. 7 is a diagram illustrating a third embodiment of a method according to the invention, and FIG. 8 is a diagram illustrating a fourth embodiment of a method

according to the invention.

  Illustrated in FIG. 2 an automatic retransmission method associated with a

  resequencing and spacing method according to the invention.

  The left vertical line E in FIG. 2 is a diagram showing, on the transmitter side, the transmission of packets as a function of time (time flow towards the

7 2781630

  bottom) and the straight vertical line R is a diagram showing, on the receiver side, the reception of packets transmitted also as a function of time. As for the central part L between the two lines E and R, it represents the transmission of packets on the transmission link. The packets which are sent by the transmitter on the link here belong to two virtual connections VC, and VCp. The number of virtual connections is given as

  example but, as will be understood later, is indifferent to the invention.

  Each packet is for example sent by the transmitter with two serial numbers: its rank in the sequence of packets which are sent on the transmission link and its rank in the sequence of packets which belong to the same connection. The first number relates only to the automatic packet repetition mechanism and, this mechanism being foreign to the present invention, it will not be considered subsequently. As for the second, it will be noted, for a packet i of a given virtual connection,

  SNi and will be called rank in the sequence, implying the connection considered.

  In Fig. 2, we also see two memories Ma and Mo respectively associated with the virtual connections VC, and VCp. For each memory Ma, Mo, each box is associated with a packet i of the associated connection and contains the rank in the sequence SNik of said packet i of the connection k and a programming time Tik, k being an index listing the connection. As will be seen below, this programming time Tk corresponds to the time when the packet i of the connection k will be delivered, by a spacer device SCH, on an output multiplex S. Thus is allocated to each packet i of rank SNi a programming time Ti. As will be understood hereinafter, the programming time Tik is determined as a function of the programming times already allocated to the packets of the same connection k of ranks lower than the packet i and so that said programming times of the packets of the

  connection are spaced from each other by at least a predetermined duration.

  Note that the number of memories is equal to the number of virtual connections supplying the transmitter. Furthermore, each memory Ma, Mp comprises as much

  of boxes that there are packets in a sequence of the corresponding connection.

  Once it has been assigned to a packet i of order SNik of a connection k for a programming time Tik, this packet is transmitted to a spacing device SCH which is generally called 'scheduler' in the field of technical. Such a device ensures that the packet i in question will be transmitted on the output multiplex S at a time which will either be the programming time Tik, or a time close to the time of

8 2781630

  k programming Tk so that a minimum time spacing t between packets not necessarily belonging to the same connection is respected. Such a device can be considered as an array in which are stored the packets waiting to be delivered on the output multiplex S. At all times t, a pointer moves from box to box of the table. If in the box indicated, a packet is stored, it is delivered on the output multiplex S. In this FIG. 2, it can be seen that the ac packet is sent on the transmission link and has been received correctly, on the receiver side. It is associated with the first box in the Ma memory corresponding to its VC <connection. In this case, the rank of the packet just received SNI 'and the programming time Tia le are also stored.

  concerning. The a2 packet was not received correctly by the receiver.

  The packet f31 which belongs to the virtual connection VCp is associated with the first box of the memory Mp corresponding to its connection VCp o its rank SN1I and its

  Ti programming times are stored.

  The packet a3 received correctly is associated with the third box of the memory Ma

  o its rank SN3 "and its programming time T3 'relating to it are then stored.

  In addition, on receipt of the packet f13, this not being in the order of the packets as they were sent on the link, the receiver deduces that the packet a2 has not been received correctly. It then transmits a negative acknowledgment (not shown) to the transmitter which, later, then retransmits it. The re-transmitted packet is then received correctly and it is then associated with the second box in the memory Ma o le rang

  SN2 'and the programming time T2 <relating to it are also stored.

  Finally, the packet P32 which belongs to the virtual connection VCp is associated with the second box of the memory Mp corresponding to its connection VCp o rank SN2a and

  the T2u programming time relating to it are stored.

  Concerning the connection VCa, the packet a is delivered, by the device SCH, to the output multiplex S at the end of its programming time Ti '. It is the same for the packets a2 and a3 which, at the end of the programming times T2 'and T3', are delivered, by the device SCH, on the output multiplex S. Similarly, concerning the connection VCp, the P3 and 32 packets are delivered by the SCH device on the output multiplex S at the end of the corresponding programming times TP and T2p. It can be seen that the packets on the multiplex S are, on the one hand, resequenced, that is to say that they are found in the order of their respective ranks in the sequence

9 2781630

  of the corresponding connection and, on the other hand, can be correctly spaced, depending on the programming times stored in each memory box. In addition, by the presence of the spacer device SCH, the packets which do not necessarily belong to the same connection are spaced at least from the interpacket time t. It can be seen that the processing of one connection is independent of that of another. Thus, the number of connections processed is therefore indifferent to the method of

  the invention. Also, in the rest of the description, we can consider only one

  connection and the index k affecting the rank in the sequence and the programming time

  of a packet of a connection k will be omitted to simplify the writing.

  The remainder of the description relates to the determination of the programming time.

  Ti which is assigned to a packet i of rank SNi in the sequence of packets of a virtual connection and which is stored in the box of the memory M associated with said connection, said box itself being associated with packet i. This package i of rank SNi

  is received on the receiver side at the time noted ti.

  We will note thereafter T the minimum time between two successive packets of a

same connection.

  Furthermore, we will define a time interval noted which is the time interval from the time of the first transmission of a packet which has not been received correctly during which said packet can be re-transmitted. We deduce from this interval 8, the maximum time for each packet beyond which, if it is received, it will no longer be

  consider valid and will therefore be destroyed.

  We will, first of all, consider that the traffic concerned by the connection considered is of the CBR (Constant Bit Rate) type, that is to say at

constant flow.

  Then according to the invention, the programming time Ti of the packet just received i is equal to the programming time Ti allocated to the first packet 1 of the sequence of packets of the same connection added to a number of intervals of interpacket time T equal to the difference of the ranks SNi and SNI respectively of said packet just received i and of said first packet. The programming time Ti is therefore given by the following relation:

2781630

  o, we recall, SNi and SNI are respectively the respective ranks of

packages i and 1.

  The first packet I of the sequence of packets of the same connection is advantageously programmed for the programming time Tl corresponding to its reception time tl to which the maximum waiting time 6 (Tl = t1 + 6) is added. For example, if we consider the example of the three packages in FIG. 3 transmitted on the transmitter side respectively at times T, 2T and 3T, the packet 2 being retransmitted at time 5T, in the memory M associated with the connection considered there will be, for a maximum waiting time equal to 3.5T: SNi Ti

1 5T

2 6T

3 7T

  It can be seen that there is, on the one hand, resequencing and, on the other hand, spacing with a time between programming time equal to the minimum time between two successive packets of the same connection or interpacket time interval T. On will now consider the case where the traffic on the considered connection is at

  variable bit rate (VBR = Variable Bit Rate = variable bit rate).

  Recall that the time Ti is the programming time of the packet i of rank SNi in the sequence of the connection considered, packet i received on the receiver side at the time noted ti. Besides the rank SNi and the programming time Ti, a flag si is also stored in the memory M associated with the connection considered which indicates whether the programming time stored Ti in the memory box associated with the packet i can be

  changed or not. For example, in the following description, we will consider that if this

  flag is at 1, it can no longer be modified and conversely, if it is at 0, it can be changed.

  Recall that we denote by T the inter-packet time interval and 6 the time

maximum wait.

  The process of the invention is illustrated by the flow diagram of FIG. 4.

  This method consists, in a first step 10, on receiving the first packet I of rank SNI = 1 in the sequence of packets of a given connection, to initialize the programming time Tl of this packet, for example at its time of it 2781630 reception t, and to position the flag El of this packet 1 so that it indicates that

  the programming time Ti allocated to this packet can no longer be modified.

  Step 20 consists of a waiting loop for the next packet of just higher rank. Step 30, which is implemented on receipt of a newly received packet from the connection, consists in determining the rank SNi of the packet i just received, in assigning to said order packet i of order SNi its programming time Ti then in store the latter in the memory box M. The value of its flag if is set to 1 so that it

  indicates that the programming time Ti of the packet just received i cannot be modified.

  According to the invention, the programming time Ti allocated to the just received packet i is determined as follows. If the flag if., Of the packet of the same connection of rank SNi-l just lower than the rank SNi of the packet just received i (SNi-l = SNi - 1) is 1, this means that it has been correctly received and that it has acquired a Ti. programming date, which can no longer be modified. Consequently, for the spacing between the packets i-I and i of respective rows SNi. and SNi is at least equal to the inter-packet time interval, the programming time Ti of the packet i is the sum of the programming time Ti-, the preceding packet and the inter-packet time T, unless the date of reception t; of the package i is later than this programming time Ti. In the latter case, the programming time Ti is equal to the reception time t ;. We can therefore write: Ti = max (Ti, + T, ti) if EF, = 1 On the other hand, if the flag si- of the packet of rank SNi-1 just below the rank SNi of the packet just received is zero, this means that the previous packet of rank SNi.4 was not received correctly and that therefore the just received packet i is not in

  sequence order. It is also said that there is an imbalance.

  We can see that if there had been no disequencing, the previous packet of rank SNi-l should have arrived at the latest a time T (T interpackage time) before the reception time t; of the packet just received i, that is, by definition, at the latest time ti - T. However, according to the definition of the maximum waiting time 5, the reception of the previous packet of rank SNil following its retransmission (s) should intervene at the latest at this time ti - T plus the maximum waiting time 5, ie ti - T +. So that the spacing between the packets i-1 and i of respective rows SN1, and SN, is at least

12 2781630

  equal to the inter-packet time interval, the programming time Ti of the packet i must therefore be this time ti - T + ô added to the inter-packet time T (ti - T + 8 + T), ie

  the sum of the reception time ti and the maximum time S waiting for the retransmission.

  We therefore have: Ti = ti + ô if Sl = O The two above expressions can be rewritten into one which takes into account the value of the flag s, of the preceding packet of rank just lower SNi: Ti = max {Ti- i + T, ti + (1c-l)}

  The package just received i is delivered to the spacing device SCH so that it

  ci then delivers it on the output multiplex S at the programming time Ti or at

  a time close to the programming time Ti.

  In Fig. 5, another embodiment of a method according to the invention has been illustrated, which consists, in addition to the preceding steps 10, 20 and 30 which are identical, of implementing a step 40 in which, for all the packets j of the same connection as the packet just received i whose rank SNj is lower than the rank SNi of the packet i just received and for which the flag sj is equal to 0 (the packet j is not yet received so that the value of programming time Tj can still be modified), is allocated a programming time Tj equal to the programming time Ti of the packet just received i from which is subtracted a number of interpacket time intervals T equal to the number of packets between the packet i and the package j, that is: Tj = Ti + (SNj-SNi) T if sj = 0 This date Tj can no longer be modified thereafter. In this case, we position

  the flag zj to the value which indicates this state, that is to say 1.

  According to another characteristic of the invention, this operation will be limited to a number of preceding packets W of rank lower than the rank SNi of the packet just received i, this number of packets W being the maximum number of packets on the connection during the maximum time standby S. So these are the packages which have not yet

  received but which, if they are, will be taken into account.

13 2781630

  We can write: 6 = WT + s with ú <T For these W packets, step 40 therefore consists in implementing the following relation: pourj = iW to it, Tj = Ti + (SNj-SNi) T sij = 0 For each of these W packets, the Esj flag is set to one, since their

  respective programming times can no longer be changed.

  On receipt of a packet i of rank SNi, two cases can occur. In the first, the reception time ti of this packet is earlier than the programming time Ti already allocated to said packet i. Then the expression giving the programming time Ti of the packet just received i is not modified and the packet i is transmitted to the device

SCH spacing for date Ti.

  In the second case, the reception time ti of the just received packet i is later than the programming time Ti already allocated to said packet i. So we consider that the

  package just received i arrived too late and it is destroyed.

  We will now consider, by way of example, a sequence of 8 packets of which

  the rows are respectively 1 to 4 and 6 to 9, the packet of row 5 not having been received.

  The reception times ti are those indicated in FIG. 6.

  The programming time Ti of the packet of rank 1 received at time t1 = T is equal to this time T so that it is delivered immediately. Since this moment of

  programming cannot be modified, the sl flag is set to 1.

  The memory cells M therefore take the following form: SNi ci Ti

I 1I T

  Packet 2 is received at time t2 = 2T. Since the flag if of the packet 1 of rank just lower is equal to 1, the programming time T2 of the packet of rank 2 is

  max (t2, TI + T) or, in this case 2T. Flag 62 of pack 2 is set to 1.

  The memory cells M then take the following form:

14 2781630

SNi _i Ti

1 1 T

2 2T

  Packet 4 is received at time t4 = 4T. Packet 3 has therefore not been received but can still be. Since the flag E3 of the packet of rank 3 just lower than the rank 4 of the packet just received is equal to 0, the programming time T4 of the packet 4 is t4 + soit or, in this case 7.5T. The programming time T3 of the rank 3 packet, which has not been received, is determined so that the packet of rank 4 is respected. I is therefore programmed for the time 6.5T. The flags of the packets of order 3 and 4 are positioned at 1I The boxes of the memory M are then presented in the following form: SNi ci i T 1 1 Ti

2:21

3 1 6.5T

4 1 7.5T

  Packet 6 is received at time t6 = 6T. The next lower tier 5 package has not yet been received, but can still be. Since the flag s5 of the rank 5 packet is

  equal to 0, the programming time of packet 6 is t6 + ô, that is 9.5T in this case.

  The programming time of the rank 5 packet, which has not been received, is determined so that the spacing with the just received packet of rank 6 is respected. This programming time is therefore time 8.5T. The flags s5 and e of the rank 5 packages

and 6 are positioned at 1.

  The memory cells M then take the following form:

2781630

SNi Si Ti

1 1 T

2 1 2T

3 1 6.5T

4 I 7.5T

1 8.5T

6 1 9.5T

  At time, t3 = 6.7T, the rank 3 packet is received. However, time t3 is later

  at programming time T3 and packet 3 is therefore destroyed.

  It will be understood that at the end of the reception of the packet of order 7, the boxes of the memory M are in the following form: SNi Ei Ti

1 1 T

2 1 21

3 1 6.5T

4 1 7.5T

1 8.5T

6 1 9.5T

7 1 10.5T

  Indeed, the flag of the previous pack of order 6 is equal to 1. The time of

  programming of the packet of order 7 is equal to max (t7, T6 + T), i.e. 10.5T.

  Another embodiment of the method of the invention is illustrated in FIG. 7.

  This method consists, in addition to steps 10 to 40 previously described, of implementing a step 50 in which, to each of the packets j of the same connection and of rank SNj greater than the rank SNi of the packet just received i, is assigned a fictitious programming time Tj which is equal to the programming time Ti allocated to the just received packet i to which is added a number of interpacket time intervals T equal to the number of packets between the packet i and the concerned packet j, ie: Tj = Ti + (SNj- SNi) T

16 2781630

  According to another characteristic of the invention, this operation will be limited to W packets subsequent to the packet just received i: pourj = i + l to i + W, Ti = Ti + (ji) T For each of these packets, the flag ej stays at zero, since it has not been

  yet received and that their fictitious programming time is not yet final.

  We will consider, as an example, the sequence of the 8 packets of rank from I to 4

and 6 to 9 of FIG. 6.

  The programming time T] of the packet of rank I received at time t = T is equal to this time T so that it is immediately delivered to the output multiplex S. Since this programming time Ti cannot be modified, we set our flag

CI to 1.

  Future packets of ranks SNj = 2 to I + W greater than rank I of the packet just received, is assigned the programming times Tj = Ti + (SNi - 1) T, while

now their flag at 0.

  The boxes of the memory M therefore have the following form (W = 3): SNi | e, Ti 1 1 _o 2T

2 0 2T

3 0 3T

4 | 0 4T

  Packet 2 is received at time t2 = 2T. Since the flag E. of the packet of rank 1 just lower is equal to 1, the programming time T2 of the packet 2 is max (t2, T2) or, in this case 2T. The flag E2 of package 2 is set to 1. In the future package

  of rank 5, the programming time T5 = 2T + 3 x T = 5T is allocated.

  The memory cells M then take the following form:

17 2781630

SNi Ei Ti

1 I T

2 2T

3 0 3T

4 0 4T

0 5T

  Packet 4 is received at time t4 = 4T. The packet of rank 3 just lower has therefore not been received but can still be. Since the flag E3 of the previous packet is equal to 0, the programming time of packet 4 is max (t4 + 5, T4), that is to say 7.5T. The flags of the packets of ranks 5, 6 and 7 are positioned at 0, so that their programming times are already scheduled at T4 + (SNi - SNi) T, ie 6.5T, 8.5T, 9 , 5T and 10.5T. The flags of packages of order 3 and 4 are

  positioned at I while those of future packets are left at zero.

  The memory cells M are then presented in the following form: SNi E _ i Ti

I 1 T

2 I 2T

__3 1 6.5T

4 1 7.5T

0 8.5T

6 0 9.5T

7 | 0 | 10.5T

  Packet 6 is received at time t6 = 6T. Package 5 has not been received but can still be. Since the flag E5 of the previous packet is equal to 0, the programming time of packet 6 is max (t6 + b, T6), that is to say 9.5T. The flags of packets 3 and 4 are already one, and therefore the programming times T3 and T4 are no longer changed. As for the flags of the packets of rank 5, 7, 8 and 9, they are all equal to 0, so that their programming times are already scheduled at T6 + (SNi- SN6) T, i.e. 8.5T , 10.5T 11.5T and 12.5T. The

18 2781630

  flags of rank 5 and 6 packets are set to 1 while those of future ones

packets are left at zero.

  The memory cells M are then presented in the following form: SNi E; Ti

1 1 T

2 I 2T

3 1 6.5T

4 1 7.5T

I 8.5T

6 1 9.5T

7 0 10.5T

8 | 0 11.5T

9 [0 12.5T

  At time, t3 = 6.7T, the packet of order 3 is received. However, the time t3 is later than the programming time T3 and, since the flag E3 is equal to 1, the packet 3 is destroyed. It will be understood that at the end of the reception of the packet of order 7, the boxes of the memory M are in the following form:

19 2781630

SN; i Ti

_ _ 1 _ _ T

2 1_ 2T

3 1 6.5T

* 4 1 7.5T

I I 8.5T

6 1 9.5T

7 1 10.5T

8 0 11.5T

9 0 12.5T

0 13.5T

  Indeed, the flag of the previous packet of order 6 is equal to 1. It follows that the programming time T7 of the packet of rank 7 is given by the relation T7 =

  Max (t7, T7), that is to say T7 previously calculated equal to 10.5T.

  Illustrated in FIG. 8 another embodiment of a method of the invention which further consists in verifying that the rank SNi of the packet i received is not greater than the rank SNm of the highest packets of the same connection already received to which is added the value W of the number of packets likely to be received during the maximum waiting time 5. If this is the case, it is also necessary to ensure a sufficient space of time between the packet of rank SNm highest and the packet just received i of rank SNi so that, in this space, the W packets preceding the packet just received i can be taken into account. Is then assigned to the packet just received i, a programming time Ti which is the highest value between the programming time calculated as above, that is to say t, + 5, and the programming time T ,, , assigned to the packet m of highest SNm rank added to a time space equal to W + I interpacket time intervals T. In addition, any packet whose rank is lower than the rank SNm of the packet m

  to which we have subtracted W which arrives is destroyed.

  To do this, a step 60 of the method consists in memorizing the highest rank SNm of the packets hitherto received. A step 70 also consists in comparing the rank SNi of the packet just received i with the highest rank SNm of the packets hitherto received to which is added W. In the case where the rank SNi is greater than the highest rank SNm of the packets hitherto received to which W is added, the method of the invention consists in calculating the programming time Ti of the packet i in the following manner:

2781630

Ti = max (ti + 5, Tm + (W + 1) T).

  We will describe the course of this embodiment by means of the following example where a packet 1 is received at a time T, the packet 2 is received at a time 5T and the

packet 3 is received at 8T time.

  As before, the programming of packet 1 is done as follows SCNi Si Ti

-1 1 T

  Packet 2 is received at time t2 = 5T and its programming time T5 is

also at 5T.

  Packet 3 is received at time t3 = 8T. Its programming time is

max (t3, T2 + T), i.e. 8T.

So we have the following situation:

SCN. T,

_ _ _ _ _ _ _ _ _ i T i

1 1 T

2 1 5T

3 1 8T

  Packet 8 is received at time ts = 10T. It will be noted that the rank 8 of this packet 8 is greater than the rank of the packet of highest rank hitherto received increased

of W (8> 3 + 3).

  We can also see that the packet of order 4 will no longer be taken into consideration and if it is received, it will be destroyed. As for the W packets 5, 6 and 7 preceding the packet just received 8, they can still be received. Consequently, to ensure the spacing between them, the packet 8 must be allocated a time interval spaced from the programming time T3 (the highest order packet which has already been received) of (W + 1) T. The programming time Tg of packet 8 will therefore be

  Max (T3 + (W + 1) T, T8 + b), i.e. here 14T.

21 2781630

  2! After applying the method of the invention, the boxes of the memory M then take the following form: SCNi Ei Ti

1 1 T

2 I 5T

3 1 8T

4 0 0

1I liT

6 1 12T

_ _ _ _ _ _7 __ _ _ _ I 13T

8 1 14T

9 0 15T

0 16T

1 1 0 17T

  In the description which has just been made, it has been considered that the maximum time

  waiting 6 is invariable, for example, for the duration of the call. Gold,

  there are circumstances where this time 6 can advantageously be modified.

  This can be the case when the packet of rank SNi just below the rank SNi of the packet just received has not yet been received correctly and would no longer be taken into account if it arrived, because the maximum waiting time 6 is expires. In this case, there is no need to delay the programming of the packet of rank SNi from b, since the packet of SNijl can no longer arrive. The programming time Ti of the packet of rank SN, is equal to the longest time between, on the one hand, the reception time ti of the packet just received i and, on the other hand, the programming time Ti which has already has been assigned to the packet of rank SNi, that is: Ti = max {Ti, ti} To do this, each packet of rank SNi of the connection concerned contains a signaling bit BSi which indicates whether the packet can still arrive (for example

  BSi = 0) or if it can no longer arrive (BSi = 1).

22 2781630

  When a packet of rank SNi arrives, the signaling flag BSi is tested. If it is zero, indicating that the packet of rank SNi can still arrive, the programming time Ti is determined as above. On the other hand, if it is one, indicating that the packet of rank SNi.1 will no longer be taken into account if it arrives, the programming time Ti is determined by the above relationship. It will be noted that this relation is similar to that of the general case but in

  considering a maximum waiting time 6 'equal to 6.BSi.l.

  Another case could be one where we can know the number r of re-

  packet emissions. If we consider a time t, between two retransmissions, the maximum waiting time 6 for a re-transmitted packet r times is advantageously equal to: 6 '= 6 - r. t, Thus the programming time of the packet of rank SNi is equal to: Ti = max {Ti., + T, ti + (1 - I) '} To be able to implement this characteristic, each packet must include a field in which is indicated the number of times it has been re-issued (at the

first emission, we have r = 0).

23 2781630

Claims (14)

  1) Method for resequencing and spacing packets of at least one connection transmitted by a transmitter and received by a receiver in view of their delivery on an output multiplex, each packet being assigned a rank in the sequence of packets which are transmitted by said transmitter and which belong to the same connection, characterized in that it consists in allocating to each packet received by said receiver a programming time corresponding to the time at which it will be delivered on said output multiplex and determined according to programming times allocated to packets of the same connection of ranks lower than its own and so that said   programming times are spaced from each other by at least a predetermined duration.   2) Method according to claim 1, characterized in that it consists, if the traffic on the connection to which belongs said just packet received by said receiver is at constant speed, to allocate a programming time equal to the programming time allocated to first packet in the sequence of packets from the same connection added to a number of inter-packet time intervals equal to the rank difference   1 5 of said just received packet and of the first packet of the packet sequence.   3) Method according to claim 2, characterized in that it consists in assigning to the first packet of the sequence of the packets which are transmitted by said transmitter and which belong to the same connection a programming time which is equal to its   reception time by said receiver plus the maximum waiting time.   4) Method according to claim 1, characterized in that it consists, if the traffic on the connection to which said just received packet belongs is at variable speed, - if the packet of the same connection of rank just lower than said just received packet has already been received, to be allocated to said just received packet a programming time equal to the programming time already determined for said packet of just lower rank added to a time equal to an inter-packet time interval, unless the date of reception of said just received packet is later than this programming time and, in this case, to allocate to said just received packet a programming time equal to this reception time, and - if the packet of the same connection of rank just lower than said just packet received has not yet been received, to be allocated to said just received packet a programming time equal to the time of reception of said just received packet added to a predetermined time. 24 2781630   ) Method according to claim 1, characterized in that it consists, if the traffic on the connection to which said just received packet belongs is of variable speed, - if the packet of the same connection of rank just lower than said just received packet has already received, to allocate to said packet a programming time equal to the longest time between, on the one hand, the programming time already allocated to said packet and, on the other hand, the time of reception of said packet, - if the packet from the same connection of rank just below said just received packet has not yet been received, to allocate to said just received packet a programming time equal to its reception time added to a predetermined delay, then to assign to each of the packets of the same connection of higher rank to said just received packet a fictitious programming time which is equal to the programming time allocated to said just received packet to which is added a number of inter inter-packet times equal to the number of packets between the packet just received and the package concerned.   6) Method according to claim 1, characterized in that it consists, if the traffic on the connection to which belongs said just received packet is at variable speed, - if the packet of the same connection of rank just lower than said just received packet has already been received, to allocate to said packet a programming time equal to the longest time between, on the one hand, the programming time already allocated to said packet and, on the other hand, the time of reception of said packet, - if the packet of the same connection of rank just below said just received packet has not yet been received, to allocate to said just received packet a programming time equal to its reception time added to a predetermined delay, then to assign to each packet a number of packets of ranks greater than the just received packet equal to the number of packets likely to be received during the maximum waiting time, a fictitious programming time equal to the program time mation of said just received packet to which is added a number of inter-packet time intervals   equal to the number of packets between the packet just received and the packet concerned.   7) Method according to claim 4, S or 6, characterized in that it consists in assigning to the first packet of the sequence of the packets which are transmitted by said transmitter and which belong to the same connection a programming time which is equal to   its time of reception by said receiver. 2781630   8) Method according to one of claims 4 to 7, characterized in that said delay   corresponds to the maximum time during which the packet can be retransmitted at the transmitter.   9) Method according to one of claims 4 to 8, characterized in that said delay   is canceled if the packet from the same connection as said packet just received but of rank   just lower will no longer be taken into account if received.   ) Method according to one of claims 4 to 9, characterized in that said delay   is reduced by a quantity depending on the number of retransmissions to which it has been subject   said package just received before receipt.   11) Method according to one of claims 4 to 10, characterized in that it consists of   allocate to all the packets of the same connection of lower rank to said just received packet and to which a definitive programming time has not been allocated, a fictitious programming time which is equal to the programming time of said just received packet to which is subtracted a number of inter-packet time intervals equal to the number of packets between the packet just received and the packet concerned, and, when the date of reception of a packet just received is later than said already allocated fictitious programming time   said package, to destroy said package just received.   12) Method according to one of claims 4 to 11, characterized in that it consists of   allocate to packets of the same connection of lower rank to said just received packet, which have not yet been received and which can still be received and taken into account a fictitious programming time which is equal to the programming time of said just received packet at which is subtracted a number of inter-packet time intervals equal to the number of packets between the packet just received and the packet concerned and, when the date of reception of a packet just received is later than said notional programming time   already assigned to said packet, to destroy said packet just received.   13) Method according to one of claims 10 to 12, characterized in that it consists,   if the rank of the packet just received is greater than the highest rank of packets from the same connection already received by a number corresponding to the number of packets likely to be received during the maximum waiting time, to be allocated to said packet just received a programming time equal to the longest time between, on the one hand, the reception time added to a predetermined time corresponding to the maximum waiting time and, on the other hand, the programming time allocated to the packet whose rank is the highest of the packets from the same connection received so far summed 26 2781630   a number of inter-packet time intervals equal to the number of packets likely   to be received during the maximum waiting time plus one unit.   14) Method according to one of claims 4 to 13, characterized in that it consists of   memorizing each packet rank number received by said receiver in the box of a memory associated with the connection to which said packet belongs, said box being associated with the rank of said packet in the sequence of packets which are transmitted by said transmitter and which belong at the same connection, to memorize in said box the programming time allocated to said packet, then to deliver said packet to a spacer.   15) Method according to claim 14, characterized in that it consists in associating with each box of said memory a flag indicating whether the programming time   associated with said box can still be changed or not.   16) Method according to claim 14 or 15, characterized in that it consists in positioning a signaling bit in each packet which indicates whether said packet will be or not taken into consideration.