JP4252596B2 - Packet transfer device - Google Patents

Description

  The present invention relates to a packet transfer device, a packet transfer network system, and a terminal device that transfer packets, and more specifically, a packet transfer device that transfers packets using forward error correction, a packet transfer network system, and The present invention relates to a terminal device.

  The technologies for realizing data communication such as files that require reliability can be broadly classified. An automatic repeat request (ARQ: Automatic Repeat reQuest) method that requests retransmission when there is an error in received data, and data There are two methods: a forward error correction (FEC) method that performs error correction when there is an error.

  A typical example of the ARQ method is TCP (Transmission Control Protocol). TCP ensures reliability by retransmitting a packet when packet discard occurs. TCP also has functions such as window control, slow start algorithm, and high-speed retransmission algorithm to avoid congestion and suppress packet discard.

  On the other hand, the FEC method is a method in which an original packet composed of original data to be transmitted and a redundant packet for error correction are transmitted together. For example, the transmitting side generates n−k (k <n, n is a natural number) redundant packets from k (1 <k, k is a natural number) original packets in the original packet, and generates n pieces of redundant packets. Send the packet. If the receiving side can receive k packets out of the transmitted n packets, it can recover the discarded packets from the received k packets, so that a retransmission request can be made as in the ARQ scheme. There is no need to do.

  In Patent Document 1, the receiving terminal measures the number of discarded packets, feeds back the measured result to the transmitting terminal, and the number of redundant packets in the FEC scheme (redundancy) based on the number of discarded packets fed back by the transmitting terminal. ), And a technique related to a data distribution control method adapted to an optimum redundancy is disclosed.

JP 2002-330118 A

  However, in the prior art described in Patent Document 1, since feedback is performed between End and End (between the transmission terminal and the reception terminal), the transmission terminal is used in a wide area network (WAN) or the like having a large network scale. In the situation where the delay time is long when a packet is transmitted to a receiving terminal via a plurality of data transfer apparatuses, there is a problem that it is not possible to sufficiently cope with a rapidly changing network state.

  Also, in a network with a long distance and a large bandwidth, for example, LFN (Long Fat Network), when the network quality is not good, control for avoiding congestion is performed and the bandwidth is not fully used (performance is not achieved). )Sometimes. In order to improve such a problem, it is conceivable that the FEC method is applied to the lower layer of TCP to lower the apparent discard rate.

  However, in this case, the packets discarded by the FEC method are restored in spite of the occurrence of congestion. Therefore, the packets are not discarded when viewed from TCP (determined that no congestion has occurred). Congestion control will not work. For this reason, there is a problem that a normal TCP to which the FEC method is not applied is adversely affected.

  In addition, since congestion is detected only when the degree of congestion is higher (the level of congestion at which packets cannot be restored by the FEC method), the performance is worsened by applying the FEC method to the lower layer of TCP. There was also a problem that occurred.

  The present invention has been made in view of the above. A packet transfer apparatus, a packet transfer network system, and a packet transfer network system that optimizes the redundancy of the FEC scheme and realizes highly reliable communication even in a large-scale network such as a WAN, and the like. A first object is to obtain a terminal device.

  A second object is to obtain a packet transfer device, a packet transfer network system, and a terminal device that can perform congestion control without degrading network performance even when the FEC method is used for a lower layer of TCP. That is.

  In the present invention, a redundant packet for forward error correction is generated based on the original packet and the redundant packet received from the transfer source via the network, and the generated redundant packet and the original packet are transferred via the network. In the packet transfer apparatus to be transmitted first, when performing decoding and encoding for the forward error correction, the redundancy is determined based on feedback information from the transfer destination, and the original received from the transfer source A forward error correction processing unit that generates feedback information to be notified to the transfer source according to the reception state of the packet and the redundant packet is provided.

  According to the present invention, when decoding and encoding for forward error correction are performed, the redundancy is determined based on feedback information from the transfer destination, and the original packet and the redundant packet received from the transfer source are received. Since feedback information to be notified to the transfer source is generated according to the state, there is an effect that highly reliable packet transfer can be performed with optimum redundancy even in a network with a long delay time such as WAN. .

  Hereinafter, embodiments of a packet transfer device, a packet transfer network system, and a terminal device according to the present invention will be described in detail with reference to the drawings. In addition, this invention is not limited by this embodiment.

Embodiment 1 FIG.
Embodiment 1 of the present invention will be described with reference to FIGS. FIG. 1 is a conceptual diagram showing an example of a network configuration to which the packet transfer apparatus according to the first embodiment of the present invention is applied. The network to which the packet transfer apparatus according to the first embodiment of the present invention is applied includes terminals 11 and 13 and a plurality of (in this case, three) FEC decoding / encoding (packet restoration / redundant packet generation). Packet transfer apparatuses (hereinafter referred to as FEC nodes) 12a, 12b, and 12c to be performed, and intermediate networks 14a and 14b are provided.

  First, taking as an example a case in which a terminal 11 (hereinafter referred to as a transmitting terminal 11) transmits a packet to a terminal 13 (hereinafter referred to as a receiving terminal 13), the FEC node of the first embodiment is applied. An overview of the network will be described.

  When transmitting the packet, the transmission terminal 11 determines the redundancy based on the feedback information 16a from the FEC node 12a, generates a redundant packet, and transmits the generated redundant packet together with the original packet. The FEC nodes 12a, 12b, and 12c restore the discarded original packet by FEC processing, and feedback information from the downstream (in the case of the FEC node 12a, the feedback information 16b from the FEC node 12b, and the FEC node 12b Is the feedback information 16c from the FEC node 12c, and in the case of the FEC node 12c, a redundancy packet is generated by determining the redundancy based on the feedback information 16d) from the receiving terminal 13, and the generated redundancy packet is converted into the original packet. Send with. That is, the FEC nodes 12a to 12c restore the discarded original packets, generate redundant packets with the redundancy determined based on the feedback information 16b to 16d from downstream, and combine the original packets and the redundant packets. The packet is transmitted and finally transferred to the receiving terminal 13. The receiving terminal 13 also restores the discarded original packet by FEC processing.

  Here, an example in which the FEC processing is also performed in the transmission terminal 11 and the reception terminal 13 is shown, but the transmission terminal 11 and the reception terminal 13 perform only normal transfer and only between the edge FEC node 12a and the FEC node 12c. The FEC process may be performed.

  Next, the FEC nodes 12a to 12c of the first embodiment shown in FIG. 1 will be described. The FEC nodes 12a to 12c all have the same function. The function of the FEC node will be described with reference to the block diagram showing the configuration of the FEC node 12a shown in FIG. The FEC node 12a includes a plurality of processing units that realize general functions related to packet transmission / reception. In FIG. 2, a forward error for performing FEC decoding / encoding of the FEC node 12a according to the present invention is shown. Only the components of the FEC processing unit 21 relating to correction processing (hereinafter referred to as FEC processing) are described.

  The FEC processing unit 21 is input from a syndrome processing unit 22 that performs FEC syndrome calculation, a restoration processing unit 23 that restores discarded packets, a syndrome processing unit 22, a restoration processing unit 23, and a redundancy adaptation processing unit 26. The redundant packet generation processing unit 24 that generates new redundant packets based on the information to be generated, the syndrome memory 25 that accumulates intermediate results of the syndrome calculation, and the redundancy at the time of transmission based on the feedback information 16b from the downstream FEC node 12b The determined redundancy is notified to the syndrome processing unit 22 and the redundant packet generation processing unit 24, and information such as the packet discard rate from the syndrome processing unit 22 is sent to the upstream node (in the case of the FEC node 12a, the transmission terminal). 11 is an upstream node and FEC node 12b Node 12a is the upstream node, and in the case of FEC node 12c, FEC node 12b is the upstream node), redundancy adapting processing unit 26 for notifying the packet flow, flow identifying unit 27 for identifying the packet flow, and received original packet A selector 28 is provided that selects one packet from three types of packets, ie, the restored original packet and the redundant packet, and sequentially transmits the selected packet.

  Next, the operation of the FEC processing unit 21 of the FEC node according to the first embodiment of the present invention will be described using the FEC node 12a as an example. When the feedback information 16b is received from the downstream FEC node 12b, the redundancy adaptation processing unit 26 determines the redundancy based on the feedback information 16b (step S100). The redundancy adaptation processing unit 26 notifies the determined redundancy to the syndrome processing unit 22 and the redundant packet generation processing unit 24.

  When the packet is received, the flow identification unit 27 determines whether or not the received packet is an FEC flow packet (step S101). If the received packet is not an FEC flow packet, the received packet is transmitted via the selector 28 without performing processing related to FEC.

  When the received packet is an FEC flow packet, the flow identification unit 27 accesses the syndrome memory 25, and the syndrome memory 25 that has received the access outputs the result of the stored syndrome calculation to the syndrome processing unit 22. (Step S102).

  The syndrome processing unit 22 calculates the syndrome from the syndrome calculation result from the syndrome memory 25 and the received packet, and stores the calculation result in the syndrome memory 25 (steps S103 and S104).

  The syndrome processing unit 22 stores the sequence number of the received packet for each flow (step S105).

  It is determined whether or not the packet received by the flow identification unit 27 is an FEC flow packet. If the packet is an FEC flow packet, the syndrome processing unit 22 calculates the syndrome and the final packet is received. (Steps S101 to S106).

  When the syndrome calculation is completed after receiving the final packet, the syndrome processing unit 22 extracts the sequence number missing from the stored packet sequence number, that is, the sequence number of the discarded packet (step S107). The syndrome processing unit 22 outputs the calculated syndrome value and the sequence number of the discarded packet to the restoration processing unit 23 and the redundant packet generation processing unit 24, calculates the packet discard rate, and outputs the packet discard rate to the redundancy adaptation processing unit 26. Output.

  The restoration processing unit 23 restores the discarded original packet based on the syndrome calculation value and the sequence number of the discarded packet (step S108). The restoration processing unit 23 outputs the restored original packet to the redundant packet generation processing unit 24 and the selector 28.

  The redundant packet generation processing unit 24 generates a redundant packet based on the syndrome calculation value, the discarded sequence number, and the restored original packet (step S109). The redundant packet generation processing unit 24 outputs the generated redundant packet to the selector 28.

  The selector 28 determines whether the input packet is the received original packet (original packet that has not been discarded), the original packet restored by the restoration processing unit 23, or the redundancy generated by the redundant packet generation processing unit 24. The packet is identified, and the received original packet, the restored original packet, and the redundant packet are transmitted in this order.

  When the discarded packet discard rate is input, the redundancy adaptation processing unit 26 generates feedback information 16a based on the packet discard rate input from the syndrome processing unit 22, and the generated feedback information 16a is downstream ( In this case, it transmits to the transmission terminal 11) (step S110).

  Next, specific examples of packet restoration and redundancy change will be described with reference to FIG. In FIG. 4, when four original packets O # 1 to O # 4 and two redundant packets R # 1 and R # 2 are to be input to the FEC node 12a, the original packet O # 3 is discarded. Then, three original packets O # 1, O # 2, O # 4 and two redundant packets R # 1, R # 2 are input. That is, the FEC node 12a receives the encoded block (the received packets of the same flow, in this case, five packets) from which the original packet O # 3 with the sequence number 3 is discarded.

  The original packets O # 1, O # 2, and O # 4 with sequence numbers 1, 2, and 4 are subjected to syndrome calculation by the syndrome processing unit 22, and immediately after that, from the desired port via the selector 28, the FEC node 12b. Sent to.

  After receiving the redundant packet R # 2 which is the final packet, the original packet O # 3 with the sequence number 3 restored by the restoration processing unit 23 is transmitted from the desired port to the FEC node 12b via the selector 28. Subsequently, new redundant packets R # 1 'to R # 3' generated by the redundant packet generation processing unit 24 are transmitted from the desired port to the FEC node 12b via the selector 28. Although the order of the original packets is changed, the receiving terminal 13 may correct the reversal of the packet order in the middle by finally performing the rearrangement of the original packets simultaneously with the restoration of the original packets.

  As described above, in the first embodiment, the feedback information is not notified between the transmission terminal 11 and the reception terminal 13, but the FEC nodes 12a to 12c arranged between the transmission terminal 11 and the reception terminal 13 are self-nodes. The feedback information is notified to the upstream node, the redundancy is determined based on the feedback information notified from the downstream node of the own node, and the FEC nodes 12a to 12c restore the original packets discarded respectively. Therefore, even in a network having a long delay time such as a WAN, highly reliable packet transfer can be performed with an optimum redundancy.

  Further, when the FEC function installed in the terminal is simply applied to an intermediate node, in order to restore the packet, the packet of the encoded block (for example, the original packet O # 1, O # 2, O in FIG. Since all of # 4, R # 1, and R # 2) must be accumulated, a large amount of memory capacity is required for packet restoration processing, and it is difficult to handle many flows due to memory limitations. However, in the first embodiment of the present invention, since only the number of redundant packets (intermediate result of syndrome calculation) is stored in the syndrome memory, the original packet discarded with a small amount of memory can be restored, More flows can be handled with the same memory capacity than when the FEC function installed in the terminal is simply applied.

Embodiment 2. FIG.
A second embodiment of the present invention will be described with reference to FIGS. In the first embodiment, the encoding itself (redundant packet) is changed to match the redundancy, but in this second embodiment, the encoding (redundant packet) is the same, and the redundancy is reduced by discarding or restoring the packet. To change.

  Since the network to which the packet transfer according to the second embodiment of the present invention is applied is the same as the network to which the packet transfer apparatus according to the first embodiment shown in FIG. 1 is applied, the description thereof is omitted here.

  FIG. 5 is a block diagram showing a configuration of the FEC processing unit 21a included in the FEC nodes 12a to 12c according to the second embodiment of the present invention. The FEC processing unit 21a according to the second embodiment of the present invention shown in FIG. 5 replaces the redundancy adaptation processing unit 26 of the FEC processing unit 21 according to the first embodiment shown in FIG. 26a, and a redundancy adjustment unit 29 is provided instead of the redundant packet generation processing unit 24. Components having the same functions as those of the FEC processing unit 21 of the first embodiment shown in FIG. 2 are denoted by the same reference numerals, and redundant description is omitted.

  The redundancy adaptation processing unit 26a determines the redundancy at the time of transmission based on feedback information from the downstream, notifies the determined redundancy to the syndrome processing unit 22, and the packet discard rate from the syndrome processing unit 22, etc. In addition to the function of transmitting the above information as feedback information upstream, the received coding block redundancy (the value obtained by subtracting the number of discarded packets from the coding redundancy) and the redundancy requested from downstream Based on the above, it is determined whether to output a processing request to the restoration processing unit 23 or to output a processing request to the redundancy adjusting unit 29.

  FIG. 6 shows the relationship between processing requests to the restoration processing unit 23 and the redundancy adjustment unit 29 based on the received redundancy of the encoded block and the required redundancy. As shown in FIG. 6, when the redundancy Ri of the encoded block that can be received is larger than the redundancy Ro requested from the downstream, that is, the number of packets that can be received more than the number of packets requested downstream. If there are many, the redundancy is excessive, and it is not necessary to restore the discarded original packet, and it is necessary to discard the packet in accordance with the requested redundancy Ro. Therefore, the redundancy adaptation processing unit 26a does not output a processing request to the restoration processing unit 23, and the number of packets to be transmitted to the redundancy adjustment unit 29 becomes the requested redundancy Ro. Output a processing request to discard the packet.

  When the redundancy Ri of the encoded block that can be received and the requested redundancy Ro are equal, that is, when the number of packets requested downstream is equal to the number of packets that can be received, the original packet needs to be restored. However, it is not necessary to discard the packet, and the received packet may be transmitted as it is. Therefore, the redundancy adaptation processing unit 26 a does not output a processing request to the restoration processing unit 23 and the redundancy adjustment unit 29.

  When the redundancy Ri of the encoded block that can be received is smaller than the redundancy Ro requested from the downstream, that is, when the number of packets that can be received is smaller than the number of packets requested downstream, the redundancy is increased. Since there is not enough, it is necessary to restore the discarded original packet. Therefore, the redundancy adaptation processing unit 26a outputs a processing request for restoring the discarded original packet so that the number of packets to be transmitted to the restoration processing unit 23 becomes the requested redundancy Ro.

  Next, the operation of the FEC processing unit 21a of the FEC node according to the second embodiment of the present invention will be described by taking the FEC node 12a as an example with reference to the flowcharts of FIGS.

  When the feedback information 16b is received from the downstream FEC node 12b, the redundancy adaptation processing unit 26a determines the redundancy based on the feedback information 16b (step S200 in FIG. 7). The redundancy adaptation processing unit 26a notifies the syndrome processing unit 22 of the determined redundancy.

  When the packet is received, the flow identification unit 27 determines whether or not the received packet is an FEC flow packet (step S201 in FIG. 7). If the received packet is not an FEC flow packet, the received packet is transmitted via the selector 28 without performing processing related to FEC.

  When the received packet is an FEC flow packet, the flow identification unit 27 accesses the syndrome memory 25, and the syndrome memory 25 that has received the access outputs the result of the stored syndrome calculation to the syndrome processing unit 22. (Step S202 in FIG. 7).

  The syndrome processing unit 22 calculates the syndrome from the syndrome calculation result from the syndrome memory 25 and the received packet, and stores the calculation result in the syndrome memory 25 (steps S203 and S204 in FIG. 7).

  The syndrome processing unit 22 stores the sequence number of the received packet for each flow (step S205 in FIG. 7).

  It is determined whether or not the packet received by the flow identification unit 27 is an FEC flow packet. If the packet is an FEC flow packet, the syndrome processing unit 22 calculates the syndrome and the final packet is received. (Steps S201 to S206 in FIG. 7) are repeated.

  When the syndrome calculation is completed after receiving the final packet, the syndrome processing unit 22 extracts the sequence number missing from the stored packet sequence number, that is, the sequence number of the discarded packet (step S207 in FIG. 7). ). The syndrome processing unit 22 outputs the calculated syndrome value and the sequence number of the discarded packet to the restoration processing unit 23, calculates the packet discard rate, and outputs the packet discard rate to the redundancy adaptation processing unit 26a.

  The redundancy adaptation processing unit 26a compares the redundancy Ri of the encoded block that can be received by subtracting the number of discarded packets from the redundancy of encoding and the redundancy Ro requested from the downstream. When the redundancy Ri of the encoded block that can be received is larger than the redundancy Ro requested from the downstream (step S208 in FIG. 8, Yes), the redundancy adaptation processing unit 26a determines the number of packets to be transmitted. The processing request including the number of packets to be discarded so as to match the requested redundancy Ro is output to the redundancy adjustment unit 29. The redundancy adjusting unit 29 discards the number of packets included in the processing request (step S209).

  When the redundancy Ri of the received encoded block is smaller than the requested redundancy Ro (step S210 in FIG. 8, Yes), the redundancy adaptation processing unit 26a requests the number of packets to be transmitted. The processing request including the number of original packets to be restored so as to be matched with the redundancy degree Ro is output to the restoration processing unit 23. The restoration processing unit 23 restores the number of original packets included in the processing request based on the syndrome calculation value and the sequence number of the discarded packet (step S211 in FIG. 8).

  When the number of discarded packets is input, the redundancy adaptation processing unit 26a generates feedback information 16a based on the packet discard rate input from the syndrome processing unit 22, and downstream the generated feedback information 16a ( In this case, the data is transmitted to the transmission terminal 11) (step S212 in FIG. 8).

  As described above, in the second embodiment, instead of newly generating a redundant packet, the original packet discarded is restored based on the redundancy Ri of the received encoded block and the requested redundancy Ro. Or, by discarding packets and adjusting the number of packets to be transmitted, the optimal redundancy can be achieved even in a network with a long delay time, such as WAN, by performing only the original packet restoration process. Can perform highly reliable packet transfer.

Embodiment 3 FIG.
A third embodiment of the present invention will be described with reference to FIGS. In the third embodiment, a redundant packet generation processing unit of an FEC node that generates redundant packets first will be described.

  FIG. 9 is a block diagram showing a configuration of the redundant packet generation processing unit 24 shown in FIG. The redundant packet generation processing unit 24 includes an encoder 241 for generating a redundant packet based on the redundancy at the time of transmission obtained by the redundancy adapting processing unit 26 based on feedback information from the downstream, and reception of the original packet. A timer 242 for notifying the encoder of the end of encoding when the interval is measured and becomes equal to or greater than a preset threshold value is provided.

  Next, referring to the flowchart of FIG. 10, based on feedback information from downstream, n (k <n, n is a natural number) −k pieces from k (1 <k, k is a natural number) original packets. The operation of the redundant packet generation processing unit of the third embodiment will be described by taking as an example a case where n redundant packets are generated and n packets are generated.

  When the original packet is received, the encoder 241 performs an encoding process for generating a redundant packet (steps S300 and S301). After performing the encoding process, the encoder 241 outputs the original packet to the selector 28 (step S302).

  Each time the encoder 241 outputs the original packet to the selector 28, the encoder 241 counts the number of output original packets, and determines whether or not the output original packet is the k-th original packet after starting the encoding ( Step S303). If the output original packet is not the k-th original packet, the encoder 241 resets the timer 242 (step S305), restarts the measurement of the reception interval of the timer 242, and then uses the received original packet. The operation of outputting the original packet after performing the encoding process is repeated until the kth packet is output (steps S300 to S304).

  When the output original packet is the k-th original packet after starting the encoding, the encoder 241 stops the timer 242, and then encodes the first original packet with the k-th original packet. Based on the above, a redundant packet is generated (step S305), and the generated redundant packet is output to the selector 28.

  On the other hand, when there is no reception of the original packet and there is a timeout notification from the timer 242, that is, when the original packet is not received even if the reception interval measured by the timer 242 exceeds the threshold value (step S 306). The encoder 241 generates a redundant packet based on the result of encoding with the original packet received before receiving the timeout notification (step S305), and outputs the generated redundant packet to the selector 28.

  As described above, in the third embodiment, when a redundant packet is first generated, the redundant packet generation processing unit 24 is provided with the timer 242 to measure the reception interval of the original packet, and a time longer than the threshold has elapsed. Since the redundant packet is generated based on the result of the encoding process using the original packet received so far, for example, even if the amount of information to be transmitted (number of original packets) is unknown, Can be generated.

  In the third embodiment, a method of transmitting nk redundant packets even when the number of received original packets is less than k has been described. However, nk packets are matched with the redundancy requested by the feedback information. Some redundant packets may be discarded and output.

Embodiment 4 FIG.
A fourth embodiment of the present invention will be described with reference to FIGS. In the third embodiment, the redundant packet generation processing unit 24 of the FEC node that performs encoding first has been described. In the fourth embodiment, the intermediate FEC node, that is, the FEC node that receives the redundant packet, A case where a packet including a packet is discarded will be described.

  In the FEC node according to the fourth embodiment of the present invention, the syndrome processing unit 22 according to the first embodiment shown in FIG. 2 is configured as shown in FIG. The syndrome processing unit 22 of the FEC node according to the fourth embodiment of the present invention includes a syndrome calculation unit 221 that calculates a syndrome from a received packet, and a threshold at which the measured reception interval is set by measuring the packet reception interval. In such a case, a timer 222 that notifies the syndrome calculation unit 221 of the completion of the syndrome calculation is provided.

  The operation of the syndrome processing unit 22 of the FEC node according to the fourth embodiment of the present invention will be described with reference to the flowchart of FIG. 12 and FIG. When receiving the packet, the syndrome calculation unit 221 resets the timer 222 and restarts the measurement of the reception interval of the timer 222 (steps S400 and S401).

  The syndrome is calculated from the intermediate result of the syndrome calculation held in the syndrome memory 25 and the received packet, and the calculation result is held in the syndrome memory 25 (steps S402 and S403).

  The syndrome calculation unit 221 stores the sequence number of the received packet for each flow (step S404).

  When the syndrome calculation unit 221 receives the packet, the syndrome calculation unit 221 resets the timer 222 and repeats the operation of calculating the syndrome until the final packet is received (steps S400 to S405).

  When the final packet is received, the syndrome calculation unit 221 stops the timer 222 and then searches for the sequence number missing from the stored sequence number, that is, the sequence number of the discarded packet, and discards the sequence of the discarded packet. A number is extracted (step S406).

  On the other hand, if no packet is received and a timeout notification is received from the timer 222, that is, if no packet is received even if the reception interval measured by the timer 222 exceeds the threshold value (step S407), storage is performed. The sequence number missing from the sequence number, that is, the sequence number of the discarded packet is searched to extract the sequence number of the discarded packet (step S406).

  The restoration processing unit 23 restores the original packet with the extracted sequence number, and the redundant packet generation processing unit 24 generates a redundant packet based on the received original packet and the restored original packet.

  The relationship between the measurement time of the timer 222 and the received packet will be described with reference to FIG. As shown in FIG. 13, four original packets O # 1 to O # 4 and three redundant packets R # 1 to R # 3 are transmitted from the transmission terminal 11 as encoded blocks. Assume that the original packet O # 4 and redundant packets R # 2 and R # 3 are discarded. That is, assume that redundant packet R # 3, which is the last packet, is discarded.

  The timer 222 is reset when the original packet O # 3 is received, and restarts the measurement of the reception interval. Although the original packet O # 4 has been discarded, the redundant packet R # 1 is received next, so the timer 222 is reset when the redundant packet R # 1 is received, and restarts the measurement of the reception interval. However, since the redundant packets R # 2 and R # 3 to be received next to the redundant packet R # 1 are both discarded, the packet is not received after the redundant packet R # 1. Therefore, the reception interval of the timer 222 exceeds the threshold value, and a timeout notification is output to the syndrome calculation unit 221. Thereby, even when R # 3 which is the final packet cannot be received, the syndrome calculation unit 221 can end the calculation of the syndrome according to the reception interval.

  As described above, in the fourth embodiment, the syndrome processing unit 22 includes the timer 222 to measure the packet reception interval, and when the measured packet reception interval exceeds the threshold value, Since it is determined that the packet has been discarded and the packet is restored and the redundant packet is generated based on the syndrome calculation value of the packet received so far, even if the final packet is discarded, the discarded original packet Restoration processing and re-encoding can be performed.

Embodiment 5 FIG.
Embodiment 5 of the present invention will be described with reference to FIGS. FIG. 14 shows the timing of received packets and transmitted packets in the FEC processing unit 21 of the FEC node of the first embodiment shown in FIG. In FIG. 14, the reception is encoded (10, 6) (6 original packets are O # 1 to O # 6, 4 redundant packets are R # 1 to R # 4). The timing when packet O # 4 and redundant packets R # 2 and R # 3 are discarded is input to the FEC node and re-encoded and transmitted at (9, 6) is shown.

  As described above, the FEC processing unit 21 according to the first embodiment receives and transmits packets while performing syndrome calculation, and after receiving all the packets that are not discarded, restores the discarded original packets. ing. That is, as shown in FIG. 14, during the period from the reception of the original packet O # 1 that is the first packet to the reception of the redundant packet R # 4 that is the final packet, the syndrome processing unit 22 Each time O # 1-O # 3, O # 5, O # 6 and redundant packets R # 1, R # 4 are received, syndrome calculation is performed. Then, after the syndrome calculation of the redundant packet R # 4 is completed, the original packet O # 4 discarded by the restoration processing unit 23 is restored. After the original packet O # 4 is restored, the redundant packet generation processing unit 24 Redundant packets R # 1 'to R # 3' are generated. For this reason, there is a problem that it takes time until the restored original packet O # 4 is transmitted after the last received original packet O # 6 is transmitted.

  In the fifth embodiment, in order to improve such a problem, when the number of packets necessary for restoring the discarded original packet is received, the restoration of the discarded original packet is started, and finally, This shortens the time from transmission of the received original packet to transmission of the restored original packet.

  FIG. 15 is a block diagram showing a configuration of the FEC processing unit 21b included in the FEC nodes 12a to 12c according to the fifth embodiment of the present invention. The FEC processing unit 21b according to the fifth embodiment of the present invention counts the number of packets received for each coding block (flow) to the FEC processing unit 21 according to the first embodiment shown in FIG. Has been added. Components having the same functions as those of the FEC processing unit 21 of the first embodiment shown in FIG. 2 are denoted by the same reference numerals, and redundant description is omitted.

  Next, the operation of the FEC processing unit 21b according to the fifth embodiment of the present invention will be described with reference to FIG. When the syndrome processing unit 22 of the FEC processing unit 21 according to the first embodiment receives the final packet and completes the syndrome calculation, the sequence number of the discarded packet is extracted from the stored sequence number, and the extracted discard is performed. The sequence number and syndrome calculation value of the received packet are output to the restoration processing unit 23 and the redundant packet generation processing unit 24 (see steps S106 and S107 in FIG. 3), whereas the FEC processing unit 21b The counter 30 counts the number of packets received for each coding block, and when the count value reaches the value of the number of packets that can be restored, the start of the restoration process is notified to the syndrome processing unit 22 Sequence number stored immediately before the syndrome processing unit 22 receiving the notification The sequence number of the discarded packet is extracted from the packet, and the extracted sequence number of the discarded packet and the calculated syndrome value are output to the restoration processing unit 23 and the redundant packet generation processing unit 24. Since this is the same as the FEC process shown in the flowchart of FIG. 3, detailed description thereof is omitted here.

  In FIG. 16, when the original packet O # 1, which is the first received packet, is received, the received packet counter 30 starts counting the number of received packets, and the syndrome processing unit 22 starts syndrome calculation.

  Packets are received in the order of original packets O # 2, O # 3, O # 5, and O # 6, and the syndrome processing unit 22 receives the original packets O # 2, O # 3, O # 5, and O # 6. Syndrome calculation is performed each time.

  When the redundant packet R # 1 is received, the count value of the reception packet counter 30 becomes “6”, and when the number of packets that can be recovered is the minimum number of packets that can be recovered, the reception packet counter 30 The start is notified to the syndrome processing unit 22. When receiving the notification of the start of the restoration process, the syndrome processing unit 22 extracts the sequence number of the discarded packet from the sequence numbers stored immediately before, and extracts the sequence number of the discarded packet and the calculated syndrome value. Is output to the restoration processing unit 23 and the redundant packet generation processing unit 24. Thereby, as shown in FIG. 16, after receiving the redundant packet R # 1, the restoration processing unit 23 starts the restoration processing of the discarded original packet (in this case, the original packet O # 4). .

  As described above, in the fifth embodiment, when the received packet counter 30 counts the number of received packets of the flow and receives the minimum number of packets that can be restored, the syndrome processing unit 22 performs the restoration process. Since the restoration processing is started by requesting the start, the restored original packet is transmitted after the received original packet is transmitted, compared to the case where the restoration processing is started after receiving the final packet. Can be shortened.

  Further, when the received packet counter 30 counts the number of received packets of the flow and receives the minimum number of packets that can be recovered from the discarded original packet, the syndrome processing unit 22 is requested to start the recovery process, and the recovery process is performed. Therefore, the time for which the coding block (flow) uses the syndrome memory 25 can be shortened, and the syndrome memory 25 can be used efficiently.

However, the restoration process of the original packet is basically to solve the simultaneous linear equations. Since the simultaneous linear equations have an order of O (n ³ ) for the number of unknowns n, the calculation is greatly performed according to the number of unknowns. The amount is different. Therefore, the restoration time may be longer depending on the relationship between the packet transmission interval and the processing capability of the FEC processing means. In such a case, when a new packet is received during calculation, the content of the packet is reflected, and the unknown is reduced so that the time does not increase.

Embodiment 6 FIG.
A sixth embodiment of the present invention will be described with reference to FIGS. In the sixth embodiment, when the necessary data is already stored in the syndrome memory and the original packet restoration process or the redundant packet generation process cannot be executed, the FEC node of the FEC node is not subjected to the extra process. This reduces the processing load.

  FIG. 17 is a block diagram showing a configuration of the FEC processing unit 21c included in the FEC node according to the sixth embodiment of the present invention. The FEC node 21c according to the sixth embodiment of the present invention includes a flow identification unit 27a instead of the flow identification unit 27 of the FEC processing unit 21 according to the first embodiment shown in FIG. 2, and a selector 28a instead of the selector 28. It has. Components having the same functions as those of the FEC processing unit 21 of the first embodiment shown in FIG. 2 are denoted by the same reference numerals, and redundant description is omitted.

  In addition to the flow identification unit 27 of the first embodiment, the flow identification unit 27a monitors the use state of the syndrome memory 25 and restores the original packet in the restoration processing unit 23 and the redundant packet generation process in the syndrome memory 25. It further has a function of determining whether there is an unused area that can be executed by the redundant packet generation process in the unit 24 and determining whether the restoration process and the redundant packet generation process can be executed. The flow identification unit 27 stores, for each flow, whether or not restoration processing or redundant packet processing has been performed as processing determination information.

  The selector 28a selects and outputs the received original packet, the restored original packet, or the redundant packet based on the processing determination information stored in the flow identification unit 27.

  Next, the operation of the FEC processing unit 21c of the sixth embodiment will be described with reference to the flowchart of FIG. When there is a usable area in the syndrome memory 25, operations for performing the syndrome calculation process, the restoration process, and the redundant packet generation process are the same as the FEC process in the flowchart shown in FIG. Therefore, the description is omitted here.

  The flow identification unit 27a monitors the usage status of the syndrome memory 25 and holds process determination information indicating whether restoration processing and redundant packet generation processing have been performed for each flow received.

  When the packet is transmitted, the selector 28a determines whether the syndrome memory has no free space (the unused area is insufficient) based on the process identification information, and whether the restoration process or the redundant packet generation process is not performed. (Step S500). When the packet to be transmitted is a packet of a flow for which the restoration process or the redundant packet generation process has not been executed, the selector 28a transmits the received packet as it is (step S501).

  If the packet to be transmitted is a packet of a flow that has undergone restoration processing and redundant packet generation processing, the selector 28a determines whether the received packet is an original packet (step S502). If the received packet is an original packet, the selector 28a transmits the received packet as it is (step S501).

  When there is no received packet, or when the received packet is not an original packet, the selector 28a determines whether there is a packet (restored original packet) that can be transmitted to the restoration processing unit 23 (step S503). . When there is a packet that can be transmitted to the restoration processing unit 23, the selector 28a transmits the packet that can be sent by the restoration processing unit 23 (step S504).

  When there is no packet that can be transmitted to the restoration processing unit 23, the selector 28a transmits a packet (redundant packet) that can be transmitted by the redundant packet generation processing unit 24 (step S505).

  As described above, in the sixth embodiment, since the unused area of the syndrome memory 25 is small, the packet received from the upstream FEC node is not used for the packet of the flow in which the restoration process or the redundant packet generation process cannot be executed. Is transmitted to the downstream FEC node as it is, so that even if the original packet discarded at the local node cannot be restored, the restoration process can be performed at the downstream FEC node. That is, since it is possible to restore original packets that are distributed and discarded throughout the network, many flows can be handled.

Embodiment 7 FIG.
A seventh embodiment of the present invention will be described with reference to FIG. First, the operation of the FEC node according to the seventh embodiment of the present invention will be described. FIG. 19 is a diagram showing the relationship between the received packet and the transmitted packet of the FEC node 12a according to the seventh embodiment of the present invention. In FIG. 19, reception is encoded (6, 5) (five original packets are O # 1 to O # 5, one redundant packet is R # 1), and original packet O # 2 is encoded. , O # 4 are input to the FEC node 12a in a discarded state. In this case, since more packets than the number of redundant packets are discarded, the FEC node 12a cannot restore the discarded original packets O # 2 and O # 4. When the FEC node 12a finds that there is an irretrievable number of discards in the received packet (in this case, the original packet O # 5 is received and two packets of the original packets O # 2 and O # 4 are received) When it is determined that the packet is discarded, the syndrome process is stopped, and only the original packets O # 1, O # 3, and O # 5 are transferred and the redundant packet R # 1 is discarded.

  When the FEC node 12a has the FEC processing unit 21 shown in FIG. 2, the number of packets discarded by the flow identification unit 27 is counted, and the count value is determined from the redundancy of the own node. When the number of packets that can be discarded is exceeded, a processing stop notification is output to the syndrome processing unit 22, and only the original packet among the received packets is transmitted, and a processing change notification that discards redundant packets is sent to the selector 28. When the syndrome processing unit 22 receives the processing stop notification, the syndrome processing unit 22 stops the syndrome processing. Upon receiving the processing change notification, the selector 28 transmits only the original packet and discards the redundant packet. That's fine.

  If the FEC node 12a has the FEC processing unit 21a shown in FIG. 5, the number of packets discarded by the flow identification unit 27 is counted, and the count value is the redundancy level of the own node. When the number of packets that can be discarded is exceeded, a process stop notification is output to the syndrome processing unit 22 and the redundancy adjusting unit 29, and only the original packet is transmitted among the received packets, and the redundant packet is transmitted. A function of outputting a processing change notification to be discarded to the selector 28 is provided. The syndrome processing unit 22 stops the syndrome processing when receiving the processing stop notification, and the selector 28 transmits only the original packet when receiving the processing change notification. The redundant packet is discarded, and when the redundancy adjustment unit 29 receives the processing stop notification, the redundancy adjustment unit 29 stops the processing and receives the packet from the selector 28. It may be output.

  When the FEC node 12a has the FEC processing unit 21c shown in FIG. 17, the number of packets discarded by the flow identification unit 27a is counted, and the count value is calculated based on the redundancy of the own node. When the number of discardable packets to be determined is exceeded, a processing stop notification is output to the syndrome processing unit 22, and only the original packet among the received packets is transmitted, and a processing change notification for discarding redundant packets is selected by the selector. The syndrome processing unit 22 stops the syndrome processing upon receiving the processing stop notification, and the selector 28a transmits only the original packet and discards the redundant packet upon receiving the processing change notification. do it.

  As described above, in the seventh embodiment, it is possible to reduce the processing at each FEC node by not performing extra processing for a flow in which a number of packets that cannot be recovered has been discarded. It is possible to handle this flow.

  Note that this is indicated in the original packet that has been found to have an irreparable number of discards so that the downstream FEC node does not perform extra processing on the flows that had an unrecoverable number of discards. A flag may be set. In this case, the flow identification unit 27 identifies the flag of the received original packet, and for the original packet that has been found to have an unrecoverable number of discards, the unrecoverable number of packets has been discarded. What is necessary is just to perform the same processing as when. This eliminates the need to count the number of discarded packets, further reduces processing, and handles many flows.

Embodiment 8 FIG.
The eighth embodiment of the present invention will be described with reference to FIG. The eighth embodiment of the present invention supports discard control corresponding to a class such as Diffserv (Differentiated Services).

  When supporting discard control corresponding to a class such as Diffserv, the FEC node 12a accumulates and transmits the packet in a queue. As shown in FIG. 20 (a), two original packets O # 1 and O # 2 are stored in the queue 40. The queue 40 has a low-priority class discard threshold x1 and a high-priority class discard threshold x2 defined for discard control according to the class. The queue 40 determines whether a newly input packet is queued or discarded according to the low priority class discard threshold x1 and the high priority class discard threshold x2, the accumulated amount of packets and the class of the input packet. decide. In FIG. 20 (a), the number of packets stored in the queue 40 is two, and since the low priority class discard threshold x1 is not exceeded (congestion has not occurred), the packets are stored. The original packets O # 1 and O # 2 remain as they are.

  In FIG. 20 (b), three original packets O # 3, O # 5, O # 6 are further accumulated, and five original packets O # 1-O # 3, O # 5 are stored in the queue 40. O # 6 is accumulated and the number of packets accumulated in the queue 40 exceeds the low priority class discard threshold x1, but does not exceed the high priority class discard threshold x2. That is, clothes in the low priority class have occurred. In this case, only the low priority class packets are discarded. Therefore, if the stored original packets O # 1 to O # 5 are in the low priority class, the original packets O # 1 to O # 5 are discarded, and if they are in the high priority class, they are not discarded.

  In FIG. 20 (c), two redundant packets R # 1 and R # 2 are further accumulated, and the queue 40 has five original packets O # 1 to O # 5 and two redundant packets. A total of seven packets of R # 1 and R # 2 are accumulated, and the number of packets accumulated in the queue 40 exceeds the high priority class discard threshold x2. That is, congestion occurs in the high priority class. In this case, all packets are discarded.

  As described above, when the packet transfer network supports discard control according to a class such as Diffserv, if the original packet is mapped to the high priority class and the redundant packet is mapped to the low priority class, the original packet Disposal can be suppressed.

  The selector 28 of the FEC processor 21 shown in FIG. 2, the selector 28 of the FEC processor 21a shown in FIG. 5, the selector 28 of the FEC processor 21b shown in FIG. 15, and the selector 28a shown in FIG. However, when a packet to be transmitted is selected, a function of mapping a high priority class to the received original packet and the restored original packet and mapping a constant priority class to the redundant packet may be provided.

  As described above, in the eighth embodiment, when the packet transfer network supports the discard control according to the class, the original packet is mapped to the high priority class, and the redundant packet is mapped to the low priority. Therefore, discarding of original packets can be suppressed, restoration processing can be reduced, the load of FEC processing can be reduced, and many flows can be handled.

Embodiment 9 FIG.
Embodiment 9 of the present invention will be described with reference to FIG. FIG. 21 is a block diagram showing a configuration of the FEC processing unit 21c included in the FEC node 12a according to the ninth embodiment of the present invention. The FEC processing unit 21c included in the FEC node 12a according to the ninth embodiment of the present invention receives a monitor packet from the upstream to the FEC processing unit 21 included in the FEC node 12a according to the first embodiment shown in FIG. A receiving unit 31 and a monitor packet generating unit 32 that generates monitor packets to be transmitted downstream at regular intervals are added.

  In the first embodiment, the packet discard rate and the like are measured from the received packet (normal user packet). In the ninth embodiment, the monitor packet is used to measure the packet discard rate and the like.

  The monitor packet generator 32 of the upstream FEC node 12a generates a monitor packet and transmits it to the downstream FEC node 12b at regular time intervals. The monitor packet receiving unit 31 of the FEC node 12b calculates the number of discarded packets (discard rate) from the sequence number added to the received monitor packet. The monitor packet receiving unit 32 of the FEC node 12b notifies the redundancy adaptation processing unit 26 of the calculated packet discard rate, and the redundancy adaptation processing unit 26 generates feedback information as in the first embodiment. The generated feedback information is transmitted to the upstream FEC node 12a.

  As described above, in the ninth embodiment, monitor packets are transmitted at regular intervals to generate feedback information, so that the state of the network can be known even when user packets are not transferred. Even when user packets are transferred intermittently, the packets can be transferred with appropriate redundancy.

Embodiment 10 FIG.
A tenth embodiment of the present invention will be described with reference to FIGS. 22 and 23. In the tenth embodiment, as shown in FIG. 22, a case where the FEC node 12a copies a packet to the FEC nodes 12b and 12c and performs multicast transfer will be described.

  FIG. 23 is a block diagram showing a configuration of the FEC node 12a according to the tenth embodiment of the present invention. In addition to the FEC processing unit 21 of the first embodiment shown in FIG. 2, the FEC node 12a of the tenth embodiment of the present invention discards redundant packets in accordance with a plurality (three in this case) of redundancy. Redundancy adjustment units 33a to 33c for storing the packets and queues 34a to 34c for storing packets for each destination. Here, it is assumed that the queue 34a corresponds to the FEC node 12b and the queue 34b corresponds to the FEC node 12c.

  In FIG. 22, the FEC node 12b requests 40% redundancy as feedback information 16b from the FEC node 12a, and the FEC node 12c provides 50% redundancy as feedback information 16c to the FEC node 12a. Demands.

  Encoded blocks of three original packets O # 1 to O # 3 and one redundant packet R # 1 are input to the FEC node 12a. In the FEC node 12a, the FEC processing unit 21 of the FEC node 12c having a high redundancy requirement re-encodes based on the feedback information 16c to generate a new redundant packet. In this case, since the number of original packets is three and the redundancy requested from the FEC node 12c by the feedback information 16c is 50%, the FEC processing unit 21 of the FEC node 12 has three redundant packets. R # 1 'to R # 3' are generated. The FEC processing unit 21 sends the original packet O # 1 to the redundancy adjusting unit 33a in the preceding stage of the queue 34a corresponding to the FEC node 12b and the redundancy adjusting unit 33b in the preceding stage of the queue 34b corresponding to the FEC node 12c. -O # 3 and redundant packets R # 1'-R # 3 'are output.

  Since the redundancy requested from the FEC node 12b is 40%, the redundancy adjustment unit 33a discards one redundant packet (in this case, the redundant packet R # 2 ′) and discards the original packets O # 1-O. # 3 and redundant packets R # 1 'and R # 3' are output to the queue 34a. Since the redundancy requested from the FEC node 12c is 50%, the redundancy adjustment unit 34b causes the original packets O # 1 to O # 3 and all the redundant packets R # 1 ′ to be generated by the FEC processing unit 21 to operate. R # 3 ′ is output to the queue 34b. That is, the FEC node 12a generates a redundant packet with the highest requested redundancy among a plurality of feedback information, and then discards the redundant packet in accordance with the redundancy for each destination.

  As described above, in the tenth embodiment, when the request for redundancy is different in multicast communication, the redundant packet is not generated for each destination, but is generated in accordance with the high redundancy of the request, Since redundant packets are discarded and the individual redundancy is adjusted in accordance with the redundancy for each destination, the FEC processing can be reduced and many flows can be handled.

Embodiment 11 FIG.
Embodiment 11 of the present invention will be described with reference to FIG. FIG. 24 is a block diagram showing components related to the TCP transmission function of the terminal 11 on the transmission side and components related to the TCP reception function of the terminal 13 on the reception side according to the eleventh embodiment of the present invention. Here, in order to simplify the description, only the component part related to the transmission function is shown in the terminal 11 on the transmission side, and only the component part related to the reception function is shown in the terminal 13 on the reception side. 11 and 13 each have a transmission function and a reception function.

  The terminal 11 includes a TCP transmission processing unit 111 that performs processing related to TCP transmission, an FEC encoding unit 112 that performs FEC encoding on data from the TCP transmission processing unit 111 and generates redundant packets, 3 includes a layer 1-3 processing unit 113 that performs the third transmission process.

  The terminal 13 includes a layer 1-3 processing unit 133 that performs layer 1-3 reception processing, an FEC decoding unit 132 that performs restoration processing of discarded packets, and a TCP reception processing unit 131 that performs processing related to TCP reception. And a window size changing unit 134 that rewrites the window size of the TCP ACK transmitted from the TCP reception processing unit 131 based on the packet discard status.

  Next, the operation of Embodiment 11 of the present invention will be described. The transmission-side TCP transmission processing unit 111 performs transmission processing related to TCP on the transmission data, and outputs the data subjected to the transmission processing to the FEC encoding unit 112. The FEC encoding unit 112 performs FEC encoding on the data that has been subjected to the transmission process, and generates a redundant packet. The layer 1-3 processing unit 113 transmits the original packet and the redundant packet as user data to the terminal 13 via the layer 1-3 processing unit 113.

  The FEC decoding unit 132 of the terminal 13 on the receiving side restores the discarded original packet based on the original packet and the redundant packet received via the layer 1-3 processing unit 133. The FEC decoding unit 132 outputs the received original packet and the restored original packet to the TCP reception processing unit 131, and stores the packet discard status (information such as the number of discards and the discard rate) for each TCP connection. The result is notified to the window size changing unit 134.

  The window size changing unit 134 rewrites the window size of the TCP ACK transmitted from the TCP reception processing unit 131 based on the notified packet discard status, and transmits it to the terminal 11 via the layer 1-3 processing unit 133. To do.

  In the terminal 11, the TCP transmission processing unit 111 changes the transmission band based on the window size of the ACK received via the layer 1-3 processing unit 113. For example, when there is no discard, the window size changing unit 134 transmits the ACK from the TCP reception processing unit 131 as it is, and decreases the window size of the ACK every time the discard increases. Thereby, the transmission band from the terminal 11 on the transmission side can be reduced.

  As described above, in the eleventh embodiment, even when the original packet is restored by FEC by applying FEC to the lower layer of TCP, the window size of ACK is reduced by the number of discarded packets. Even with low-quality LFN, performance can be extracted and congestion control can be performed appropriately, and highly reliable transfer of packets can be realized.

Embodiment 12 FIG.
A twelfth embodiment of the present invention will be described with reference to FIG. FIG. 25 is a block diagram showing structures of terminal 11 and terminal 13 according to the twelfth embodiment of the present invention. In the terminal 11 of the twelfth embodiment of the present invention, a redundancy adapting processing unit 114 is added to the terminal 11 of the eleventh embodiment shown in FIG. 24, and the terminal 13 has the implementation shown in FIG. A feedback information transmission processing unit 135 is added to the terminal 13 in the form 11.

  That is, in addition to the function of the eleventh embodiment, the feedback information transmission processing unit 135 sets the number of packets discarded and the discard rate detected by the FEC decoding unit 132 of the terminal 13 on the receiving side to the layer 1-3 processing unit 133. Feedback to the terminal 11 on the transmission side as feedback information, and based on this feedback information, the redundancy adaptation processing unit 114 of the terminal 11 determines the redundancy.

  Thus, in the twelfth embodiment, even when FEC is applied to the lower layer of TCP and the original packet is restored by FEC, the window size of ACK is reduced according to the number of discarded packets, and discarded packets are discarded. Since redundancy is determined based on feedback information such as the number of data and the discard rate, it is possible to draw out performance even with poor LFN and perform congestion control appropriately, and to suppress the occurrence of congestion And reliable transmission of packets can be realized.

Embodiment 13 FIG.
A thirteenth embodiment of the present invention will be described with reference to FIG. In the twelfth embodiment, the TCP ACK and feedback information are individually transmitted. In the thirteenth embodiment, the TCP ACK and feedback information are transmitted together.

  FIG. 26 is a block diagram showing structures of terminal 11 and terminal 13 according to the thirteenth embodiment of the present invention. In the terminal 11 of the thirteenth embodiment of the present invention, a feedback information extraction unit 115 is added to the terminal 11 of the eleventh embodiment shown in FIG. 24, and the terminal 13 has the embodiment shown in FIG. 11 includes an ACK rewriting unit 135 instead of the window size changing unit 134 of the terminal 13.

  The ACK rewriting unit 135 generates a special packet in which feedback information such as the number of discarded packets and the discard rate notified from the FEC decoding unit 132 is added to the TCP ACK.

  The feedback information extraction unit 115 extracts feedback information from a special packet including ACK and FEC feedback information. The feedback information extraction unit 115 determines the redundancy based on the extracted feedback information and notifies the determined redundancy to the FEC encoding unit 112, converts a special packet into a normal ACK format, and transmits the TCP The data is output to the processing unit 111.

  As described above, in the thirteenth embodiment, the ACK rewriting unit 135 transmits the ACK and the feedback information as one special packet, and the feedback information extraction unit 115 extracts the feedback information from the special packet. Since the format is converted to normal ACK, it is not necessary to transmit a packet to notify feedback information, and the amount of traffic on the network can be reduced.

Embodiment 14 FIG.
A fourteenth embodiment of the present invention will be described with reference to FIG. In Embodiments 11-13, the process between the terminals in TCP was demonstrated. In the fourteenth embodiment, as shown in FIG. 1, the case where the FEC nodes 12a to 12c are arranged between the terminal 11 and the terminal 13 and the redundancy is changed for each section will be described. As shown in FIG. 1, when the FEC nodes 12a to 12c are arranged between the transmission terminal 11 and the reception terminal 13, the reception terminal 13 grasps the number of discarded packets and the discard rate of the entire network. It is not possible. Therefore, the window size of ACK cannot be adjusted to an appropriate value. Therefore, an information area for recording the number of discards or the discard rate is provided in the packet to be transferred, and the reception terminal 13 is notified of the number of packets discarded and the discard rate.

  FIG. 27 is a block diagram showing a configuration of the FEC processing unit 21d included in the FEC node 12a according to the fourteenth embodiment of the present invention. In the FEC processing unit 21d according to the fourteenth embodiment of the present invention, a discard number updating unit 35 is added to the FEC processing unit 21 according to the first embodiment shown in FIG.

  The discard number update unit 35 adds the discard number of packets notified from the syndrome processing unit 22 to the discard number of the information area of the received packet, and adds the discard number of packets discarded up to the own node to the information area To do.

  As shown in FIG. 1, when a packet is transmitted from the transmission terminal 11 to the reception terminal 13, since there is no packet discarded at the transmission terminal 11, a packet with an information area of “0” is input to the FEC node 12a. Is done. When discarding two packets is detected by the syndrome processing unit 22 of the FEC node 12a, the discard number updating unit 35 of the FEC node 12a transmits a packet whose information area is “2” to the FEC node 12b. When the syndrome processing unit 22 of the FEC node 12b detects the discard of one packet, the discard number update unit 35 of the FEC node 12b detects the discard in its own node in the information area “2” of the received packet. “1” is added, and the packet with the information area set to “3” is transmitted to the FEC node 12c. When the syndrome processing unit 22 of the FEC node 12c detects no packet discard, the discard number updating unit 35 of the FEC node 12c does not change the information area (adds “0” to the information area) and does not change the information area. A packet whose area is “3” is transmitted to the receiving terminal 13. The FEC decoding unit 132 of the receiving terminal 13 extracts the number of discarded packets in the information area, and the window size changing unit 134 shown in FIG. 24 or 25 or the ACK rewriting unit 135 shown in FIG. Notify

  As described above, in the fourteenth embodiment, an information area for recording the number of discarded packets or the discard rate is provided in the packet, and the FEC node arranged for each section is configured to detect the number of discarded packets or Since the discard rate is added to the information area, the receiving terminal can grasp the number of discarded packets or the discard rate of the entire network, and the TCP ACK window size can be adjusted according to the network.

  In the fourteenth embodiment, the example in which the FEC processing unit 21 of the first embodiment shown in FIG. 2 is provided with the discard number updating unit 35 has been described. However, the FEC processing unit shown in FIG. 21a, the FEC processing unit 21b shown in FIG. 15 or the FEC processing unit 21c shown in FIG. 17 includes the discard number updating unit 35, and the FEC processing unit 21 includes the discard number updating unit 35. Needless to say, the same effect can be obtained.

  As described above, the packet transfer apparatus according to the present invention is useful for a large-scale network, and is particularly suitable for a packet transfer apparatus in a network such as a WAN having a long delay time.

FIG. 1 is a conceptual diagram showing an example of a network configuration to which the packet transfer apparatus according to the first embodiment of the present invention is applied. FIG. 2 is a block diagram showing a configuration of an FEC processing unit included in the FEC node shown in FIG. FIG. 3 is a flowchart for explaining the operation of the FEC processing unit of the FEC node according to the first embodiment of the present invention. FIG. 4 is a diagram for explaining the operation of the FEC processing unit of the FEC node according to the first embodiment of the present invention. FIG. 5 is a block diagram showing a configuration of an FEC processing unit included in the FEC node according to the second embodiment of the present invention. FIG. 6 is a diagram showing a relationship between processing requests to the restoration processing unit and the redundancy adjusting unit based on the received redundancy of the encoded block and the requested redundancy. FIG. 7 is a flowchart for explaining the operation of the FEC processing unit of the FEC node according to the second embodiment of the present invention. FIG. 8 is a flowchart for explaining the operation of the FEC processing unit of the FEC node according to the second embodiment of the present invention. FIG. 9 is a block diagram showing the configuration of the redundant packet generation processing unit of the FEC node according to the third embodiment of the present invention. FIG. 10 is a flowchart for explaining the operation of the redundant packet generation processing unit of the FEC node according to the third embodiment of the present invention. FIG. 11 is a block diagram showing the configuration of the syndrome processing unit of the FEC node according to the fourth embodiment of the present invention. FIG. 12 is a flowchart for explaining the operation of the syndrome processing unit 22 of the FEC node according to the fourth embodiment of the present invention. FIG. 13 is a diagram for explaining the relationship between the measurement time of the timer and the received packet. FIG. 14 is a diagram showing timings of received packets and transmitted packets in the FEC processing unit of the FEC node according to the first embodiment of the present invention. FIG. 15 is a block diagram showing the configuration of the FEC processing unit of the FEC node according to the fifth embodiment of the present invention. FIG. 16 is a diagram showing timings of received packets and transmitted packets in the FEC processing unit of the FEC node according to the fifth embodiment of the present invention. FIG. 17 is a block diagram showing the configuration of the FEC processing unit of the FEC node according to the sixth embodiment of the present invention. FIG. 18 is a flowchart for explaining the operation of the FEC processing unit of the FEC node according to the sixth embodiment of the present invention. FIG. 19 is a diagram showing the relationship between the received packet and the transmitted packet of the FEC node according to the seventh embodiment of the present invention. FIG. 20 is a diagram for explaining the operation of the FEC node according to the eighth embodiment of the present invention. FIG. 21 is a block diagram showing the structure of the FEC processing unit of the FEC node according to the ninth embodiment of the present invention. FIG. 22 is a conceptual diagram showing a network configuration according to the tenth embodiment of the present invention. FIG. 23 is a block diagram showing a configuration of the FEC node 12a according to the tenth embodiment of the present invention. FIG. 24 is a block diagram showing the structure of the terminal according to Embodiment 11 of the present invention. FIG. 25 is a block diagram showing the structure of the terminal according to the twelfth embodiment of the present invention. FIG. 26 is a block diagram showing the structure of the terminal according to the thirteenth embodiment of the present invention. FIG. 27 is a block diagram showing the structure of the FEC processing unit of the FEC node according to the fourteenth embodiment of the present invention.

Explanation of symbols

21 FEC Processing Unit 22 Syndrome Processing Unit 23 Restoration Processing Unit 24 Redundant Packet Generation Processing Unit 25 Syndrome Memory 26 Redundancy Adaptation Processing Unit 2
27 Flow identifier 28 Selector

Claims (3)

A packet that generates a redundant packet for forward error correction based on an original packet and a redundant packet received from a transfer source via a network, and transmits the generated redundant packet and the original packet to a transfer destination via a network In the transfer device,
When performing decoding and encoding for forward error correction, the redundancy is determined based on feedback information from the transfer destination, and the original packet received from the transfer source and the reception state of the redundant packet A forward error correction processing unit for generating feedback information to be notified to the transfer source;
With
The forward error correction processing unit is
A redundancy adaptation processing unit that determines redundancy based on feedback information from the transfer destination and generates feedback information to be notified to the transfer source based on the number of discarded original packets and redundant packets received from the transfer source; ,
A flow identification unit for identifying flows of original packets and redundant packets received from the transfer source;
A syndrome processing unit for performing syndrome calculation based on the original packet identified by the flow identification unit;
A restoration processing unit for restoring the original packet discarded by decoding;
The redundant packet is generated by re-encoding based on the syndrome calculation value calculated by the syndrome processing unit, the original packet restored by the restoration processing unit, and the redundancy determined by the redundancy adaptation processing unit A redundant packet generation processing unit,
A selector that selects and outputs the original packet identified by the flow identification unit, the original packet restored by the restoration processing unit, and the redundant packet generated by the redundant packet generation processing unit;
With
The redundant packet generation processing unit includes:
A timer that measures the reception interval of packets from the transfer source;
An encoding unit for generating a redundant packet from an original packet from a transfer source by encoding based on the redundancy determined by the redundancy adaptation processing unit;
With
When the reception interval of the original packet from the transfer source measured by the timer exceeds a predetermined threshold, a redundant packet is generated from the result of the encoder at that time. Packet transfer device. A packet that generates a redundant packet for forward error correction based on an original packet and a redundant packet received from a transfer source via a network, and transmits the generated redundant packet and the original packet to a transfer destination via a network In the transfer device,
When performing decoding and encoding for forward error correction, the redundancy is determined based on feedback information from the transfer destination, and the original packet received from the transfer source and the reception state of the redundant packet A forward error correction processing unit for generating feedback information to be notified to the transfer source;
With
The forward error correction processing unit is
A redundancy adaptation processing unit that determines redundancy based on feedback information from the transfer destination and generates feedback information to be notified to the transfer source based on the number of discarded original packets and redundant packets received from the transfer source; ,
A flow identification unit for identifying flows of original packets and redundant packets received from the transfer source;
A syndrome processing unit for performing syndrome calculation based on the original packet identified by the flow identification unit;
A restoration processing unit for restoring the original packet discarded by decoding;
The redundant packet is generated by re-encoding based on the syndrome calculation value calculated by the syndrome processing unit, the original packet restored by the restoration processing unit, and the redundancy determined by the redundancy adaptation processing unit A redundant packet generation processing unit,
A selector that selects and outputs the original packet identified by the flow identification unit, the original packet restored by the restoration processing unit, and the redundant packet generated by the redundant packet generation processing unit;
With
The syndrome processing unit
A timer that measures the interval at which packets are received from the source,
With
When the reception interval of the original packet from the transfer source measured by the timer exceeds a predetermined threshold, the restoration processing unit restores the original packet based on the syndrome calculation value at that time. A packet transfer apparatus characterized by performing the packet transfer. A packet that generates a redundant packet for forward error correction based on an original packet and a redundant packet received from a transfer source via a network, and transmits the generated redundant packet and the original packet to a transfer destination via a network In the transfer device,
When performing decoding and encoding for forward error correction, the redundancy is determined based on feedback information from the transfer destination, and the original packet received from the transfer source and the reception state of the redundant packet A forward error correction processing unit for generating feedback information to be notified to the transfer source;
With
The forward error correction processing unit is
A redundancy adaptation processing unit that determines redundancy based on feedback information from the transfer destination and generates feedback information to be notified to the transfer source based on the number of discarded original packets and redundant packets received from the transfer source; ,
A flow identification unit for identifying flows of original packets and redundant packets received from the transfer source;
A syndrome processing unit for performing syndrome calculation based on the original packet identified by the flow identification unit;
A restoration processing unit for restoring the original packet discarded by decoding;
The redundant packet is generated by re-encoding based on the syndrome calculation value calculated by the syndrome processing unit, the original packet restored by the restoration processing unit, and the redundancy determined by the redundancy adaptation processing unit A redundant packet generation processing unit,
A selector that selects and outputs the original packet identified by the flow identification unit, the original packet restored by the restoration processing unit, and the redundant packet generated by the redundant packet generation processing unit;
With
An information area for recording the number of packet discards in the original packet;
The forward error correction processing unit is
A discard number updating unit for detecting the number of discarded packets based on original packets or redundant packets received from a transfer source, and adding the detected number of discarded packets to the information area;
A packet transfer apparatus further comprising:

Images