JP4930449B2 - Forward error recovery method receiver and forward error recovery method reception output method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve sufficient resistance and low delay with respect to packet loss. <P>SOLUTION: A forward error recovery system receiving apparatus receives a stream of media packets with sequence numbers and FEC packets, and outputs the media packets. The forward error recovery system receiving apparatus includes: a buffer means which stores the received packets; a node control means which sets a plurality of node sequences comprised of a plurality of nodes for referring to storage positions, which are ranked by relatively corresponding to the sequence numbers; an analysis means which determines ranking of the nodes corresponding to the received media packets, and regards the media packets to which the nodes comprising the node sequence including the earliest ranking node need to refer when the determined ranking is within a predetermined ranking range as output objects; and an output means which attempts packet recovery using the FEC packets, and outputs the media packets to be output in the order of the sequence numbers when the media packets to be output include unreceived media packets. <P>COPYRIGHT: (C)2010,JPO&amp;INPIT

Description

  The present invention relates to a forward error recovery system receiver and a forward error recovery system reception output method, and more particularly to a forward error recovery system reception apparatus and forward error recovery system reception output capable of realizing sufficient tolerance against packet loss and low delay. Regarding the method.

  When stream data such as video / audio is transmitted in packet units using an information transmission network, some packets may be lost (packet loss). As a technique for dealing with packet loss, a packet retransmission method in which the receiving side detects packet loss and requests the transmitting side to retransmit the packet has been generally used. In recent years, attention has been paid to the Forward Error Correction (FEC) method, in which redundant information is added to information to be transmitted and transmitted, and packets lost on the receiving side are restored. ing.

For example, Patent Document 1 describes a technique for reducing video transmission delay in a video communication system using a packet loss tolerance function by FEC, and Patent Document 2 discloses a total of 100 × 10 × 10 for transferring video data. A technique for performing two-dimensional FEC with a packet is described.
JP 2007-28241 A JP 2008-11096 A

  The FEC is characterized in that a lost packet is restored on the receiving side by adding redundant information to information to be transmitted, and therefore, a procedure for requesting retransmission is unnecessary and the amount of delay can be reduced.

  In general, redundant information is generated as an FEC packet by performing a predetermined FEC operation on a plurality of packets. As a generation method thereof, a method in which a plurality of packets are FEC-calculated in one dimension, a plurality of packets in a multidimensional manner, and the like. A method for performing FEC calculation is proposed. For example, in a one-dimensional parity FEC that takes the exclusive OR (XOR) of four packets, one FEC packet is generated by four packets, whereas four packets are 2 × 2 2 In a two-dimensional parity FEC that is XORed in dimension, five FEC packets are generated by four packets.

  With regard to resistance to packet loss, for example, assuming that the probability of occurrence of packet loss is 1% at random, when 1-dimensional parity FEC is applied to four packets, packet loss occurs and packet restoration is impossible. Is 0.039%, whereas when 2 × 2 two-dimensional parity FEC is applied to four packets, the probability is 0.000004%.

  As described above, the method of performing the multi-dimensional FEC operation on the plurality of packets is overwhelmingly higher than the method of performing the FEC operation on the plurality of packets in one dimension. However, the multi-dimensional FEC calculation method for a plurality of packets not only increases the calculation amount but also increases the number of packets to be transmitted, which complicates the packet management on the receiving side and increases the delay amount. There is a problem that it ends up. In particular, if it is attempted to manage the timing at which a received packet is output to the next stage and the timing at which the packet is restored as in the prior art using the generation time, reception time, etc. for each packet, the load on the receiving side becomes enormous.

  Therefore, it is desired to develop a technique that facilitates packet management on the receiving side while maintaining resistance to packet loss. Therefore, an object of the present invention is to provide a forward error recovery system receiver and a forward error recovery system reception output method that can realize sufficient tolerance against packet loss and low delay.

  In order to solve the above-mentioned problem, a forward error recovery method receiver according to the first aspect of the present invention includes a stream of media packets assigned a sequence number and a stream of FEC packets generated based on the media packets. A forward error recovery type receiving apparatus that receives and outputs media packets in order of sequence number, for storing the received media packets and FEC packets, and for referring to the storage location of the media packets in the buffer means A node management means that sets a plurality of node sequences composed of a plurality of nodes ranked relatively corresponding to the sequence number, and determines the rank of the node corresponding to the sequence number of the received media packet; The media packet is sent to the node with the determined rank. Analyzing means for storing a delivery position and, when the determined rank is within a predetermined first rank range, a media packet to be referred to by a node constituting a node sequence including the node with the highest rank as an output target media packet And when the output target media packet includes an unreceived media packet, the stored FEC packet is used to restore the unreceived media packet into an output target media packet, and the output target media packet is sequenced Output means for outputting in numerical order.

  The received media packet is managed by the node, and the media packet is output triggered by recording to a node of a predetermined order. Therefore, time management for each packet becomes unnecessary. Therefore, even when a multi-dimensional FEC calculation method is used for a plurality of packets, the packet management of the receiving apparatus is easy, and sufficient tolerance and low delay against packet loss can be realized.

  Here, when the output means outputs the media packets to be output by the output means in the order of the sequence number, the analysis means invalidates the storage contents of the node storing the storage position of the output media packet, and You can move up the ranking. Thereby, resources can be used effectively.

  In addition, when the permutation of the determined nodes is equal to or higher than the predetermined second order, the analysis unit sets a media packet to be referred to by a node constituting the node train including the node with the highest rank as an output target media packet. It is possible to move up the ranking of subsequent nodes without any problem. This is because in the transmission of information in which real-time property is important, if a packet is received exceeding an allowable delay time, output is not necessary.

  Further, when the determined rank is in a predetermined first rank range, the analyzing means includes nodes constituting a node string having a predetermined relationship with the node string in addition to the node string including the node having the highest rank. The media packet to be referred to may be the output target media packet. This increases the degree of freedom in design.

  Specifically, the node string having the predetermined relationship may be a node string including the node having the highest rank other than the node constituting the node string including the node having the highest rank.

  In general, in consideration of compatibility, the correlation between media packets and FEC packets is described only in FEC packets. Therefore, when the FEC packet indicating the start value of the sequence number of the media packet has not been received, the analyzing means stores the storage position of the media packet received first in a predetermined node, and after receiving the FEC packet, It is desirable to determine the node order corresponding to the sequence number of the media packet, and shift the order given to the nodes so that the predetermined node has the determined node order. This facilitates the association processing between the node and the media packet when the FEC packet is first received.

  In order to solve the above problem, a forward error recovery method reception output method according to a second aspect of the present invention includes a stream of media packets assigned a sequence number, a stream of FEC packets generated based on the media packets, Is a forward error recovery method reception output method for outputting media packets in the order of sequence numbers, a buffering step for storing the received media packets and FEC packets in a buffer, and a storage position of the media packets in the buffer. A node setting step for setting a plurality of node strings composed of a plurality of nodes ranked corresponding to the sequence numbers for reference, and the ranks of the nodes corresponding to the sequence numbers of the received media packets And the node with the determined order A node storage step for storing the storage position of the media packet in the node, and a medium to be referred to by a node constituting a node sequence including the node having the earliest rank when the determined rank is in the predetermined first rank range When a packet is an output target media packet and the output target media packet includes an unreceived media packet, the stored FEC packet is used to restore the unreceived media packet into an output target media packet, An output step of outputting the output target media packets in order of sequence numbers.

  According to the present invention, it is possible to achieve both sufficient tolerance against packet loss and low delay.

  Embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a network media stream transmission / reception system including an FEC receiving side apparatus to which the present invention is applied. As shown in the figure, in the network media stream transmission / reception system 100, the FEC transmission side device 20 generates FEC media packets by a predetermined FEC operation based on the media packets sequentially output by the media stream output device 10, and the computer together with the media packets. The data is transmitted to the FEC receiving side device 30 via the network 50. The FEC receiving side device 30 outputs the received media packet to the media stream receiving device 40. At this time, the FEC receiving side apparatus 30 outputs the media packets to the media stream receiving apparatus 40 in the order of the sequence, not in the order of reception, and when some media packets are lost, the received FEC packet and others The lost media packet is restored using the media packet and output. The network media stream transmission / reception system 100 adopting such an FEC method is particularly effective for transmission of information that requires low-delay, such as VoIP (Voice over IP), TV conference, and multicast distribution system. Can be applied to.

  The media stream output from the media stream output device 10 can be various data such as moving images and sounds. Further, the media stream output device 10 and the media stream receiving device 40 may be configured to transmit and receive media streams in both directions. The computer network 50 can use a non-quality guaranteed information transmission network that performs transmission in units of packets according to a predetermined protocol, and can typically be the Internet.

  FIG. 2 is a block diagram illustrating a functional configuration of the FEC reception side device 30. The FEC reception side device 30 includes a packet reception unit 310, a node management unit 320, a packet analysis unit 330, a trigger generation unit 340, and a media packet output unit 350.

  The packet reception unit 310 includes a reception packet buffer 312 having a ring buffer structure, and sequentially stores received media packets and FEC packets in the reception packet buffer 312. Further, when receiving a packet, the packet receiving unit 310 notifies the packet analyzing unit 330 that the packet has been received. In this embodiment, in order to realize low-delay transmission, the packet stored in the reception packet buffer 312 is directly used without being copied until output to the media stream reception device 40 at the next stage. .

  In order to manage the packets stored in the reception packet buffer 312 in the reception order in the output order (sequence order), the node management unit 320 uses the node array in which the nodes indicating the packet storage positions are arranged in the output order to receive the packet buffer. 312 is referred to. That is, since the packets stored in the reception packet buffer 312 are in the order of reception, the output order may be switched or may be lost. Therefore, packets stored using nodes can be referred to in the order of output. The specific configuration and processing of the node management unit 320 will be described later.

  When the packet reception unit 310 receives a packet, the packet analysis unit 330 analyzes the header of the received packet, determines a node corresponding to the packet, and notifies the node management unit 320 of it.

  The trigger generation unit 340 generates various triggers according to the analysis result of the packet analysis unit 330. The triggers generated by the trigger generation unit 340 include a node group update / discard trigger for the node management unit 320 and a media packet output trigger for the media packet output unit 350. Conditions for the trigger generation unit 340 to generate these triggers will be described later.

  The media packet output unit 350 outputs the media packet stored in the reception packet buffer 312 to the next stage media stream reception device 40 in a predetermined unit, triggered by the output trigger generated by the trigger generation unit 340. If packet loss occurs at this time, the lost packet is restored using the packet restoration unit 352 and output.

  Here, media packets and FEC packets in the present embodiment will be described. Note that the FEC packet in this embodiment conforms to RFC2733. However, the present invention can also be applied to FEC packets having other specifications.

  FIG. 3A shows a schematic structure of a media packet. As shown in the figure, each media packet includes a header and a payload, and a sequence number is stored in the header. Note that the start value of the sequence number is generally generated by a random number.

  FIG. 3B is a diagram illustrating a schematic structure of the FEC packet. As shown in the figure, each FEC packet includes a header and a payload, and the header stores information indicating its own sequence number and the correspondence between the media packet and the FEC packet. The latter information is an example of RFC2733. Base sequence number and mask. The base sequence number is a sequence number of a media packet serving as a reference indicating the corresponding media packet. The mask is information indicating a set of media packets used to generate the FEC packet. The offset from the base sequence number of each sequence number of the media packet shown in the left is represented by a bit mask, and all of those involved in FEC packet generation are represented. It is an integer obtained by ORing the bit mask notation of the media packet. As described above, the correspondence between the media packet and the FEC packet is described in the FEC packet for the purpose of maintaining compatibility of the media stream.

  FIG. 4 is a diagram schematically illustrating a process for generating an FEC packet from a media packet and a process for restoring the media packet. FIG. 4A illustrates a case in which an FEC packet is generated with two one-dimensional elements. FIG. 4B shows a case where an FEC packet is generated with 2 × 2 two-dimensional four elements.

  As shown in FIG. 4 (a), when generating an FEC packet with two one-dimensional elements, for example, a media packet with a sequence number M1 (hereinafter, a media packet with a sequence number n is indicated as “Mn” The FEC packet of number k is indicated as “Fk”) and “M2” are used as units of FEC calculation, and “F1” is generated by calculating a predetermined operation, for example, an XOR value. Then, in addition to the media packets “M1” and “M2”, a redundant FEC packet “F1” is transmitted. In the information transmission network, for example, when “M2” is lost, the lost “M2” can be restored by calculating the XOR values of “M1” and “F1” on the receiving side.

  As shown in FIG. 4B, when an FEC packet is generated with 2 × 2 two-dimensional four elements, for example, media packets of “M1”, “M2”, “M3”, and “M4” are subjected to FEC calculation. Unit. Then, “F1” is calculated from “M1” and “M2”, “F3” is calculated from “M3” and “M4”, “F2” is calculated from “M1” and “M3”, “F4” is calculated from “M2” and “M4”. Further, “F5” is calculated from “F1” and “F3” or “F2” and “F4”. Then, the redundant FEC packets “F1”, “F2”, “F3”, “F4”, and “F5” are added to the media packets “M1”, “M2”, “M3”, and “M4”, and nine packets are transmitted.

  When an FEC packet is generated with two-dimensional four elements, in the information transmission network, for example, even when five packets of “M1”, “M2”, “M3”, “F3”, and “F4” are lost, the receiving side Then, “F3” is restored from “F1” and “F5”, “M3” is restored from the restored “F3” and “M4”, and “M1” is restored from the restored “M3” and “F2”. ”And restoring“ M2 ”from the restored“ M1 ”and“ F1 ”, so that the receiving side can obtain“ M1 ”,“ M2 ”,“ M3 ”, and“ M4 ”of the original media stream. it can.

  Next, the processing in the FEC reception side apparatus 30 having the above configuration will be described in detail with reference to a preferred embodiment.

  First, basic processing as the first embodiment will be described with reference to the flowchart of FIG. 5 and FIGS. 6 to 10 showing the configuration and processing of the node management unit 320.

  In this process, first, node and buffer are initially set (S101). In the initial setting of the node, the node management unit 320 sets the node block according to the FEC calculation method. In this example, it is assumed that a method of generating FEC packets with 2 × 2 two-dimensional four elements is used.

  In the first embodiment, a node block 321 as shown in FIG. 6A is set in the node management unit 320 in the case of generating a FEC packet with 2 × 2 two-dimensional four elements. In this figure, 12 nodes from “n1” to “n12” are set, and the contents of each node are in an unreferenced initial state. Here, for the sake of simplicity, the description will be given focusing on the media packet. For the FEC packet, a similar node block corresponding to the FEC packet may be set, or may be managed by the same node block 321 as described in a second embodiment described later.

  The node plays a role of pointer to record information for referring to the packet stored in the reception packet buffer 312, for example, an address. In this embodiment, a node constitutes one node sequence in units of FEC operations. In other words, since the FEC operation is performed using 2 media in 2 × 2 and four media packets, one node string is formed by four nodes. As a result, the node block 321 is formed with three node strings “n1” to “n4”, “n5” to “n8”, and “n9” to “n12”. The node sequence is a unit for efficiently performing processing such as output to the next stage media stream receiving device 40 and discarding.

  One or a plurality of node strings are collected to form a node string group. Three node sequence groups are configured in the node block 321 and are referred to as a first node sequence group, a second node sequence group, and a third node sequence group, respectively. Each node string group is composed of one or a plurality of node strings. In the first embodiment, one node string group is constituted by one node string. Therefore, the node strings “n1” to “n4” constitute the first node string group, the node strings “n5” to “n8” constitute the second node string group, and “n9” to “n12”. This node sequence forms the third node sequence group.

  As described above, in this embodiment, the nodes are set from “n1” to “n12”, but the node “n1” is the most sequenced media packet that has not been output to the media stream receiver 40 at the next stage. This is a node for referring to a media packet having a small number. When the corresponding media packet is stored in the reception packet buffer 312, information for specifying the storage position is recorded. Thereafter, “n2” to “n12” correspond to the sequence numbers of the media packets in the same manner. That is, referring to the reception packet buffer 312 in the order of nodes, media packets can be acquired in the order of sequences and packet loss can be detected.

  In addition, the first node column group, the second node column group, and the third node column group are configured by one or a plurality of node columns, and the first node column group is a node queue group that is waiting for output, and a second node column group. The group has a character as a node queue group waiting for reception, and the third node queue group as a preliminary node queue group.

  Since there are eight media packets corresponding to the end of the second node sequence group from the beginning of the first node sequence group, for example, if the transmission interval of media packets in the FEC transmission side device 20 is 8 ms, 8 ms × 8 = A node corresponding to a media packet corresponding to 64 ms can be recorded in the first node sequence group and the second node sequence group. Therefore, the delay time allowed in this embodiment is designed to be a maximum of 64 ms, and within this range, the received media packet is made to correspond to one of the nodes of the first node sequence group and the second node sequence group. Can do. Of the maximum delay time of 64 ms, 32 ms (8 ms × 4) is a delay time originally required for the FEC calculation.

  In the initial setting of the buffer, the packet receiving unit 310 reserves an area for the received packet buffer 312. As described above, the reception packet buffer 312 has a ring buffer structure. The size to be secured can be determined based on the configuration of the node block 321, the allowable delay amount, and the like, as will be described later.

  When the media packet is received, the received media packet is stored in the received packet buffer 312 (S102). Then, the packet receiving unit 310 notifies the packet analyzing unit 330 of reception, and the packet analyzing unit 330 determines the corresponding node N with reference to the sequence number stored in the header of the received media packet (S103). ). As described above, “n1” corresponds to a media packet that has not been output to the media stream receiver 40 in the next stage. Therefore, the sequence number of the media packet to be referred to by “n1” and the sequence number of the received media packet By referring to the above, it is possible to determine the node N corresponding to the received media packet. However, depending on the arrival order of the media packets, the corresponding node may be outside the range of “n1” to “n12”.

  In accordance with RFC 2733, the sequence number of the media packet corresponding to each FEC packet is determined from the base sequence number described in the FEC packet and the mask. Conversely, until the first FEC packet is received, the correlation between the media packet and the FEC packet is unknown, and the node where the media packet is to be stored cannot be specified. The node determination process at this time will be described later. Here, it is assumed that the FEC packet has already been received and the base sequence number has been determined.

  When the node N corresponding to the received media packet is determined, processing corresponding to N is performed (S104). First, when the determined node N is “n1” ≦ N ≦ “n6”, that is, the node before the third “n7” in the first node sequence group and the second node sequence group, the received node N The storage position of the media packet in the reception packet buffer 312 is recorded in the node N (S105). As a result, the received media packet can be indirectly referenced by referring to the node N.

  For example, when packets are received in the order of “M 1” “M 2” “F 1” “M 3” “F 2” “F 3” “F 4” “F 5” “M 6” “M 5”, the media packet received packet buffer 312 is received. And the node string are in a state as shown in FIG. Here, it is assumed that packets from the address “b1” of the reception packet buffer 312 are stored in “b2”, “b3”. As shown in this figure, “b1” storing “M1” having the smallest sequence number is recorded in “n1”, and “b2” storing “M2” is recorded in “n2”. Since the packet “M4” has not been received, the corresponding “n4” is in an unreferenced state. “M5” and “M6” are received in reverse order and are stored in the reception packet buffer 312 in the order of “M6” and “M5”, but are referred to in the order of “n5” and “n6”. Thus, it is possible to manage in sequence order.

  If the determined node N is N = “n7” or “n8”, that is, the node after the third “n7” in the second node sequence group, first, the received media packet is stored in the received packet buffer 312. The position is recorded in the node N (S106).

  Then, triggered by the reception of the media packet corresponding to the third and subsequent nodes of the second node sequence group, the output to the next stage of the first node sequence group, that is, to the media stream receiving device 40 (S107). At this time, if there is a loss in the media packet corresponding to the first node sequence group, the lost media packet is restored using the FEC packet stored in the same manner and output.

  For example, as shown in FIG. 6C, when the media packet “M7” corresponding to “n7” is received, the node indicates the media packets referred to by “n1” to “n4” in the first node sequence group. Output in order. In this figure, since “M4” that should correspond to “n4” is lost, “M4” is restored and “M1” to “M4” are output to the media stream receiving device 40 in the next stage.

  As described above, in this embodiment, reception of a packet corresponding to a predetermined node, specifically, the third and subsequent nodes of the second node sequence group is used as a trigger for outputting the media packet of the first node sequence group. For this reason, management for each packet such as transmission time, reception time, and elapsed time, which has been essential in the past, becomes unnecessary, and it is not necessary to perform periodic monitoring of received packets and timer start processing. Therefore, it is possible to reduce the processing load related to the packet management of the receiving side device and to realize a low delay.

  In the present embodiment, reception of media packets corresponding to nodes after “n7” in the second node sequence group is used as a trigger, but this is an example of setting. The media packet output trigger that minimizes the delay amount is when a media packet corresponding to “n4”, which is the last node in the first node sequence group, is received, but the media packet corresponding to “n4” is lost. It can happen if you do. Since the FEC packets “F1” to “F5” are generated from the media packets “M1” to “M4”, the “F5” of the FEC packet is used to exhibit the maximum restoration capability for the first node group. Is necessary. In addition to this, the media packet output timing should be late in order to ensure error tolerance against packet transmission delay fluctuation and order reversal. On the other hand, delaying the media packet output timing increases the amount of delay.

  If the output trigger is reception of a media packet corresponding to the k-th node, the k-th media packet already has a delay of (k−1) × 8 ms at the encoding time with respect to the top media packet “M1”. In addition, other delays are added by various processes in the FEC receiving side device 30. Therefore, the trigger node is set in consideration of the delay allowable in the media stream receiving apparatus 40 and the error tolerance.

  When the media packet corresponding to the node of the first node sequence group is output, the node sequence group is updated (S108). In updating the node sequence group, the nodes of the first node sequence group that are output and become unnecessary are invalidated as a no-reference state. As a result, the area referred to by the nodes in the first node sequence group of the received packet buffer 312 is released. The node sequence of the second node sequence group is the first node sequence group, the node sequence of the third node sequence group is the second node sequence group, and the node sequence of the unreferenced first node sequence group is the third node. Shift to each column group.

  That is, the state shown in FIG. 6C is updated to the state shown in FIG. In the example of this figure, “b10”, “b9”, and “b11” recorded in “n5”, “n6”, and “n7” of the second node column group are respectively “n1”, “n2”, and “n2” of the first node column group. n3 ". In addition, “b1”, “b2”, and “b4” that are referenced in the first node sequence group are released. Although not shown, since the same processing is performed for the FEC packet, “b3”, “b5”, “b6”, “b7”, and “b8” are also released. Thereafter, the same processing is repeated for the updated node string group while the reception is continued (S120).

  For implementation, as shown in FIG. 7B, a pointer variable group 322 indicating the first node sequence group, the second node sequence group, and the third node sequence group is prepared for updating the node sequence group. It is sufficient to exchange the values of the pointer variables. As described above, in this embodiment, the received packet buffer 312 referred to by the node string group is not referred to by the old packet due to the update process of the node string group. Therefore, the buffer release process is not required, and the size of the reception packet buffer 312 set in the process (S101) can be set to the minimum necessary size.

  If the determined node N is N <n1, that is, if the received media packet is older than the media packet corresponding to “n1”, the node N is discarded without being recorded (S109). Such a packet has already been processed as a packet loss.

  When the determined node N is “n9” or later, it cannot be recorded in the first node sequence group and the second node sequence group. Such a situation occurs when a delay exceeding the delay time of 64 ms allowed in this embodiment occurs. If the reception is within the allowable delay time, the media packet corresponding to the node of the first node sequence group is output while the media packet corresponding to the node of the second node sequence group is being received, and the node sequence group is updated. Therefore, the media packet corresponding to the node of the third node sequence group is not received.

  In this embodiment, a spare third node sequence group is provided in order to prepare for a delay greater than the allowable delay time. For the sake of convenience, the third node sequence group assigns “n9” to “n12” continuous to the second node sequence group, but has a packet reception interval and is not a continuous node to the second node sequence group. Specifically, even when a media packet corresponding to a node after “n13” is received, it can be handled safely. In the following, a description will be given of processing when a delay equal to or greater than the allowable delay time occurs, that is, when the determined node N is “n9” or later.

  If the determined node N is N = “n9” or “n10”, that is, a node before the third “n11” in the third node sequence group, first, in the received packet buffer 312 of the received media packet. The storage position is recorded in the node N (S110). Recording in the node of the third node sequence group means that the time to output the media packet corresponding to the first node sequence group to the media stream receiving device 40 in the next stage has passed. The recording contents of the nodes of the first node sequence group are discarded without outputting the media packet referred to by the nodes of the first node sequence group (S111).

  For example, as illustrated in FIG. 8A, when a media packet “M10” corresponding to “n7” is received after “M5”, the first node sequence is output without outputting “M1” to “M4”. The contents of “n1” to “n4” of the group are discarded and set to a no-reference state.

  The node sequence of the second node sequence group is the first node sequence group, the node sequence of the third node sequence group is the second node sequence group, and the node sequence of the unreferenced first node sequence group is the third node. The node column group shifted to each column group is updated (S112). By updating the node sequence group in this manner and repeating the processing for the packet received thereafter (S120), the nodes of the third node sequence group recorded in the processing (S110) are changed to the first node sequence group and the second node sequence group. It becomes possible to use the node sequence group equivalently and continuously. For this purpose, a third node column group is provided and set to the same size as the first node column group and the second node column group. In addition, this means does not provide the third node group sequence, and stores the packet analysis contents in the temporary storage area (node) and records it in an appropriate node that is vacant after updating the node sequence group. The recording process (timing) and the node reference process (timing) are unified, the competition is eliminated, the design concept is concise and clear, and the safety and reliability of the system can be ensured. FIG. 8B shows the state of the reception packet buffer 312 and the node block 321 after updating the node string group.

  When the determined node N is N = “n11” or “n12”, that is, the node after the third “n11” in the third node sequence group, first, the received media packet is stored in the received packet buffer 312. The position is recorded in the node N (S113). Recording in the node of the third node sequence group means that the time to output the media packet corresponding to the first node sequence group to the media stream receiving device 40 in the next stage has passed. The recording contents of the nodes of the first node sequence group are discarded without outputting the media packet referred to by the nodes of the first node sequence group (S114).

  However, since the media packet referred to by the node of the second node sequence group is valid within the allowable delay time, it is output to the media stream receiving device 40 at the next stage (S115). At this time, if there is a loss in the media packet corresponding to the second node sequence group, the lost media packet is restored using the FEC packet stored in the same manner and output.

  For example, as shown in FIG. 9A, when the media packet “M11” corresponding to “n11” is received after “M5”, the first node sequence is output without outputting “M1” to “M4”. Discard the contents of “n1” to “n4” of the group. In addition, media packets referred to by “n5” to “n8” of the second node sequence group are output in the order indicated by the nodes. In this figure, since “M7” and “M8” that should correspond to “n7” and “n8” have been lost, after restoring “M7” and “M8”, “M5” to “M8” The data is output to the media stream receiving device 40.

  Then, the node of the second node sequence group that has become unnecessary after output is set to the no-reference state, the node sequence of the third node sequence group is set to the first node sequence group, and the node sequence of the second node sequence group of the non-reference state is set Is updated to the third node sequence group, and the node sequence group that shifts the node sequence of the first node sequence group in the unreferenced state to the second node sequence group is updated (S116). However, the node column of the second node column group is the first node column group, the node column of the third node column group is the second node column group, and the node column of the first node column group is the third node column group, respectively. The updating of the node string group to be shifted may be repeated twice. By updating the node sequence group in this manner and repeating the processing for the subsequently received packets (S120), the nodes in the third node sequence group recorded in the processing (S113) can be used continuously. It becomes like this. FIG. 9B shows the state of the received packet buffer 312 and the node block 321 after updating the node string group. As shown in the figure, the contents of the third node sequence group are shifted to the first node sequence group.

  When the determined node N is N ≧ “n13”, that is, a node in a node sequence that is not continuous with the second node sequence group, the storage position of the received media packet in the reception packet buffer 312 is represented by Nx = n9 + ( N−n9)% 4 (% is a remainder, 4 is the number of elements in the 2D FEC of the 2 × 2 embodiment) is recorded in the node Nx of the third node sequence group (S117). That is, if the received media packet is the first media packet in the FEC operation, it is recorded in “n9”, and if it is the second media packet, it is recorded in “n10”. If it is, it is recorded in “n11”, and if it is the fourth media packet, it is recorded in “n12”.

  If the determined node N is N ≧ “n13”, the media packet corresponding to the first node sequence group and the media packet corresponding to the second node sequence group should be output to the media stream receiving device 40 in the next stage. This means that the time has elapsed, so that the media node referred to by the nodes of the first node sequence group and the second node sequence group is not output, and the first node sequence group and the second node sequence group The recorded contents of the node are discarded (S118).

  For example, as shown in FIG. 10A, when a media packet “M14” corresponding to “n14” is received after “M5”, “M14” is converted into an FEC operation performed in “M13” to “M16”. “B15” in which “M14” is stored in “n10” is recorded. Further, “n1” to “n4” of the first node sequence group and “n5” to “n8” of the second node sequence group without outputting “M1” to “M4” and “M5” to “M8”. Discard the contents of.

  The node sequence of the third node sequence group is the first node sequence group, the node sequence of the second node sequence group is the third node sequence group, and the node sequence of the first node sequence group is the second node sequence group. The node string group to be shifted is updated (S119). However, the node column of the second node column group is the first node column group, the node column of the third node column group is the second node column group, and the node column of the first node column group is the third node column group, respectively. The updating of the node string group to be shifted may be repeated twice. By updating the node sequence group in this manner and repeating the processing for the subsequent received packets (S120), the nodes of the third node sequence group recorded in the processing (S117) can be used continuously. It becomes like this. FIG. 10B shows the state of the reception packet buffer 312 and the node block 321 after updating the node string group.

  As described above, in the first embodiment, the size of the node string is made equal to the unit of the FEC operation, so that the output timing of the media packet and the update timing of the reception packet buffer 312 are associated with each other. . For this reason, the FEC receiving side apparatus 30 which has a more reliable management mechanism is realizable.

  Next, a second embodiment of the present invention will be described. In the first embodiment, for simplification, in the node block 321, one node string group is constituted by one node string. In the second embodiment, the case where one node sequence group is configured by one or a plurality of node sequences will be described more generally.

  In FIG. 11, the first node sequence group is composed of four node sequences, the second node sequence group is composed of N-4 node sequences, and the third node sequence group is composed of two node sequences. A node block 321a is shown. Also in this example, the number of nodes constituting each node row is made equal to the unit of FEC calculation.

  Here, considering the shift of the node sequence group after the node sequence is output, the number of node sequences in the first node sequence group and the number of node sequences in the second node sequence group are the third node sequence. It is an integer multiple of the number of group node sequences. That is, in this example, control is performed so that the number of node strings in the third node string group matches the number of node strings output at a time. In the case of this figure, since the number of node columns of the third node column group is two, “node column 1” and “node column 2” are simultaneously output when a predetermined condition is satisfied.

  FIG. 12 is a flowchart for explaining media packet output processing when the node block 321a as shown in FIG. 11 is used. Since the basic processing is the same as in the first embodiment, characteristic processing is shown here. In this example, it is assumed that a media packet corresponding to a predetermined node G, for example, the third and subsequent nodes in the last node sequence N of the second node sequence group is received as a trigger for outputting the media packet.

  Similar to the first embodiment, when a media packet is received (S201), it is stored in the received packet buffer 312, and the sequence number of the header is analyzed and the storage position is recorded in the corresponding node (S202). If the corresponding node is not after the node G in the second node sequence group (S203: No), the media packet is not output and the next packet is supported.

  On the other hand, if the corresponding node is after node G of the second node sequence group, the media packet corresponding to the number of nodes in the third node sequence group from the top of the first node sequence group To the next stage media stream receiver 40 (S204). At this time, if there is a packet loss, it is restored and output. In this example, “node string 1” and “node string 2” are output at a time.

  Then, the node sequence is shifted forward according to the number of output node sequences. In this example, as shown in FIG. 13, as a result of outputting “node string 1” and “node string 2”, a forward shift of two node strings is performed. That is, “node string 3” and “node string 4” included in the first node string group are shifted to the head part of the first node string group, and the “node string” included in the second node string group 5 ”and“ node string group 6 ”are shifted to the second half of the first node string group. The same shift is performed in the second node column group and the third node column group, the second node column group is composed of “node column 7” to “node column N + 2”, and the third node column group is “node column” N + 3 ”and“ node string N + 4 ”. Note that, when the packet cannot be received within the allowable delay time, the process using the third node sequence group is performed as in the first embodiment.

  In the second embodiment, it is generally assumed that the allowable delay amount is larger than that in the first embodiment. The actual allowable delay amount can be designed according to the specific configuration of the node string group and the setting of the node position that generates the output trigger. Therefore, the second embodiment has a higher degree of practical design freedom than the first embodiment. More specifically, by increasing the number of node strings constituting the second node string group, it is possible to bring the error tolerance against packet transmission delay fluctuation and order inversion to the limit of the allowable delay amount. Thus, the number of node sequences constituting the second node sequence group is a matter to be determined by the design guideline regarding the delay of the media stream receiving device 40. Further, the number of nodes of the first node sequence group is not limited to the above example, and depending on the design policy, it may be a processing unit at the time of output, a unit of FEC calculation, a unit of node sequence shift, Other processing units can be used. Further, the distinction between the first node sequence group, the second node sequence group, and the third node sequence group is convenient, for example, the third node sequence group is added to the second node sequence of the second node sequence group. By assigning a role or by assigning the role of the first node row group to the first half of the second node row group, the row groups may not be clearly distinguished.

  Next, as another example of the second embodiment, a case where different types of node columns are included in one column group will be described. The different types of node sequences can be, for example, a media packet node sequence and an FEC packet node sequence. In this example, according to the unit of FEC operation, the media packet node sequence is composed of four nodes. The node sequence is assumed to be composed of 5 node sequences. In addition, it is assumed that the media packet referred to by the media packet node sequence included in the same node sequence group and the FEC packet referred to by the FEC packet node sequence have a relationship of FEC calculation.

  For example, the media packet node sequences “n1,” “n2,” “n3,” and “n4” included in the first node sequence group of the node block 321b shown in FIG. That is, five FEC packets referred to by the FEC packet node sequence “f1”, “f2”, “f3”, “f4”, and “f5” are generated.

  In this example, the output trigger of the media packet corresponding to the first node sequence group is when the media packet corresponding to the media packet node “n7” or later of the second node sequence group is received, or the second node sequence group Suppose that the FEC packet corresponding to the FEC packet node “f7” onward is received. However, the output trigger is not limited to this example. For example, the determination criterion may be another node position or a case where both a node after a predetermined position of the media packet node and a node after a predetermined position of the FEC media packet node are received.

  FIG. 15 is a flowchart for explaining media packet output processing when the node block 321b as shown in FIG. 14 is used. Since the basic processing is the same as in the first embodiment, characteristic processing is shown here.

  As in the first embodiment, when a packet is received (S301), the packet is stored in the received packet buffer 312, and the sequence number of the header is analyzed and the storage position is recorded in the corresponding node (S302). At this time, if the received packet is a media packet, the storage location is recorded in the corresponding node of the media packet node. If the received packet is an FEC packet, the storage location is recorded in the corresponding node of the FEC packet node. To do.

  When the recorded node is after the media packet node “n7” of the second node sequence group, or when it is after the FEC packet node “f7” (S303: Yes), the media packet node sequence of the first node sequence group Are output to the next-stage media stream receiving apparatus 40 in the order of sequence numbers (S304). At this time, if there is a packet loss, it is restored using the FEC packet and output.

  Then, the same node string group as in the first embodiment is updated (S305). Specifically, the first node column group is a third node column group, the second node column group is a first node column group, and the third node column group is a second node column group. Note that, when the packet cannot be received within the allowable delay time, the process using the third node sequence group is performed as in the first embodiment.

  On the other hand, if the recorded node is not after the media packet node “n7” of the second node sequence group and after the FEC packet node “f7” (S303: No), the media packet is not output and the next is not performed. Corresponds to the packet (S301).

  Next, as another example of the second embodiment, a case where the sequence numbers of media packets used for FEC calculation are not consecutive will be described. In general, in a computer network such as the Internet, although single packet loss is dominant, it is known that when a packet is lost, the probability of loss of the packet immediately after that increases.

  As a countermeasure against such continuous packet loss, in FEC, media packets distant from each other in time series may be associated with each other and FEC calculation may be performed in a so-called interleaved format. In this example, the case where “M1”, “M3”, “M5”, and “M7” are associated with each other and “M2”, “M4”, “M6”, and “M8” are associated with each other and the FEC operation is performed in an interleave format will be described as an example.

  FIG. 16A shows an example of a node block 321c corresponding to such an interleave format. In other words, “n1”, “n3”, “n5”, “n7” constitute one node string and “n2,” “n4,” “n6,” “n8” constitute one node string corresponding to the FRC operation unit. These two node rows constitute a first node row group. For the sake of simplicity, the following description focuses on media packets. For the FEC packet, a separate node block may be provided, or two media packet node strings and two FECs including the FEC packet node string in the same node string group as shown in FIG. A node block 321d may be set by configuring one node string group with a total of four node strings in the packet node string. Each node sequence group may include three or more media packet node sequences.

  In this example, the output trigger of the media packet corresponding to the first node sequence group is assumed to be a case where a media packet corresponding to the node “n13” and subsequent nodes in the second node sequence group is received. However, the output trigger is not limited to this example. For example, the determination criterion may be another node position such as “n14”.

  When the media packet corresponding to the node serving as the output trigger is received, the media packet corresponding to the node included in the first node sequence group is output to the next-stage media stream receiving device 40 in the order of “n1” to “n8”. . At this time, if a packet loss has occurred, the packet is restored in units of FEC calculation and output. In general, when FEC is performed in an interleaved format, packet management is further complicated. However, as shown in the present embodiment, packet management can be performed very easily.

  Finally, the node setting of the media packet until the first FEC packet is received will be described. As described above, in the FEC conforming to RFC 2733, the correspondence between the sequence number of the media packet and the node is not determined until the first FEC packet is received.

  For example, as shown in FIG. 6B, when “M1” is first received, if it is not known that the media packet is the first packet, “n1” should be referred to. It is not possible to determine whether or not to refer to. Therefore, by performing the following processing, it is possible to easily associate the sequence number of the media packet with the node when the FEC packet is first received.

  First, as an initial state of the node block, the number of nodes is prepared by adding m−1, where m is the number of nodes constituting the media stream node sequence, to the number of nodes originally required. In this example, the node string is composed of four nodes, and each node string group is composed of one node string. Therefore, the number of nodes originally required is 12, and the number of extra nodes required. Becomes 3. Therefore, a total of 15 nodes are prepared as an initial state.

  As shown in FIG. 17A, the initial value of the first pointer indicating the first node sequence group is set to the mth, that is, the fourth node, and the initial value of the second pointer indicating the second node sequence group is set. The value is set to the 2mth, that is, the 8th node, and the initial value of the second pointer indicating the third node sequence group is set to the 3mth, that is, the 12th node. As a result, the first node sequence group is provisionally set at the 4th to 7th nodes, the second node sequence group is provisionally set at the 8th to 11th nodes, and the third node sequence group is set at the 12th to 15th nodes. Temporarily set.

  Assume that in this initial state, packets are received in the order of “M7”, “M8”, and “F9”. When the media packet “M7” is received, since the FEC packet has not been received, it is not possible to determine what number packet “M7” is from the start number of the sequence number. For this reason, as shown in FIG. 17B, it is made to correspond to the eighth node which is the head of the second node sequence group. Although the FEC packet is not received even when the media packet “M8” is received, it can be determined that “M8” is the next media packet after “M7”. “M8” corresponds to the ninth node.

  When the FEC packet “F9” is received, the base sequence number and the mask are acquired with reference to the header. Here, it is assumed that “M1” is found to be a packet of one row and one column in 2 × 2 two-dimensional FEC, that is, the head. Then, it can be determined that “M7” received first is a media packet that should correspond to “n8”, which is the third node in the second node sequence group. Therefore, as shown in FIG. 17C, the values of the first pointer, the second pointer, and the third pointer are shifted by minus two nodes, and the second node is the first node of the first node sequence group, that is, “n1”. ”Is set. The first, fourteenth and fifteenth nodes are not used. With the above processing, the sequence number of the media packet is associated with the node in the node block 321e.

  In the present embodiment, even when the FEC packet has not been received and the node is in a temporarily set state, the output of the media packet and the update of the node sequence group can be performed according to the flowchart shown in FIG. An example in which packets are received in the order of “M3”, “M4”, “M6”, “M7”, “M8”, and “F9” will be described with reference to FIG.

  As shown in FIG. 18A, the media packet “M3” is provisionally associated with the eighth node that is the head of the second node sequence group, “M4” is provisionally associated with the ninth node, and “M6” "Is temporarily associated with the 11th node.

  Since the eleventh node is the fourth node in the provisional second node sequence group, the provisional first node sequence group is output in accordance with the processing (S107). However, since the media packet corresponding to the node of the temporary first node sequence group has not been received, the media packet is not output after all. Next, the node string group is updated according to the process (S108). As a result, the 8th to 11th node rows become the first node row group, the 12th to 15th node rows become the second node row group, and the 4th to 7th node rows become the third node row group. Since no FEC packet has been received at this time, an FEC packet node sequence (not shown) is in a non-reference state. Next, “M7” and “M8” are received, and the state shown in FIG.

  Thereafter, when “F9” of the FEC packet is received, the sequence number of the first media packet of the media stream can be acquired. Here, it is assumed that “M1” is the head. At this time, the 8th to 11th nodes correspond to the first node sequence group, the 12th to 15th nodes correspond to the second node sequence group, and the 4th to 7th nodes correspond to the third node sequence group. Adjustment is necessary because it is temporarily set.

  The reception of the FEC packet reveals that the values of the first pointer, the second pointer, and the third pointer need only be shifted by minus two nodes. Therefore, the node block 321f is brought into a state as shown in FIG. Set. That is, the sixth node is the first node “n1” of the first node sequence group, the tenth node is the first node “n5” of the second node sequence group, and the second node is the third node sequence group. To be the first node “n9”. The first, fourteenth and fifteenth nodes are not used.

  At this stage, since the media packet corresponding to “n7” has been received, the output trigger of the first node sequence group is generated. Therefore, “M3” and “M4” referred to by “n3” and “n4” are output to the media stream receiver 40 at the next stage, and then the node sequence group is updated.

  Even if more media packets are received before the first FEC packet is received, it is not possible to restore the packet by FEC by updating the node sequence group in the same way as in the normal state according to the flowchart of FIG. The media packet can be output to the next stage. That is, an algorithm similar to that at the normal time can be used except that the association process at the first reception of the FEC packet is performed. For this reason, the handling of the media packet before receiving the FEC packet does not require a conventional packet saving resource, an algorithm separate from the normal one, and the like.

  As described above, according to the embodiment of the present invention, received packets are managed by a node sequence, and the node sequence is further managed as three node sequence groups. As a result, management on the receiving side is simple and processing is lightened, and sufficient error correction can be performed.

  In addition, by providing high-performance forward error correction (FEC) in low-delay transmission systems to restore missing data, a reception / decoding method is provided that can realize low-delay transmission over general-purpose and inexpensive transmission paths. It becomes possible to do. For example, when the present invention is applied to business broadcasting operated in various facilities, a business broadcasting system capable of low-delay and high-quality transmission using a general-purpose and inexpensive local LAN can be realized.

It is a block diagram which shows the structure of the network media stream transmission / reception system containing a FEC receiving side apparatus. It is a block diagram which shows the function structure of a FEC receiving side apparatus. It is a figure which shows schematic structure of a media packet and a FEC packet. It is a figure which shows typically the process which produces | generates a FEC packet from a media packet, and the process which restore | restores a media packet. It is a flowchart explaining the process in a FEC receiving side apparatus. It is a figure which shows the structure and process of a node management part. It is a figure which shows the structure and process of a node management part. It is a figure which shows the structure and process of a node management part. It is a figure which shows the structure and process of a node management part. It is a figure which shows the structure and process of a node management part. It is a figure which shows the example of the node block which comprised one node sequence group by the 1 or several node sequence. It is a flowchart explaining the process of the FEC receiving side apparatus in 2nd Example. It is a figure which shows the update of the node row group in 2nd Example. It is a figure which shows the example of the node block which comprised one node sequence group by the node sequence of a different kind. It is a flowchart explaining the process of the FEC receiving side apparatus in another example of 2nd Example. It is a figure which shows the example of the node block which comprised the node row group corresponding to an interleave format. It is a figure explaining the node setting of the media packet until an FEC packet is received. It is a figure explaining the node setting of the media packet until an FEC packet is received.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Media stream output apparatus, 20 ... FEC transmission side apparatus, 30 ... FEC reception side apparatus, 40 ... Media stream reception apparatus, 50 ... Computer network, 100 ... Network media stream transmission / reception system, 310 ... Packet reception part, 312 ... Reception Packet buffer, 320 ... node management unit, 321 ... node block, 322 ... pointer variable group, 330 ... packet analysis unit, 340 ... trigger generation unit, 350 ... media packet output unit, 352 ... packet restoration unit

Claims (12)

A forward error recovery method receiver that receives a stream of media packets with sequence numbers and a stream of FEC packets generated based on the media packets and outputs the media packets in the order of the sequence numbers,
Buffer means for storing the received media packet and FEC packet;
Node management means for setting a plurality of node sequences composed of a plurality of nodes ranked in correspondence with the sequence numbers for referring to the storage position of the media packet in the buffer means;
The rank of the node corresponding to the sequence number of the received media packet is determined, the storage position of the media packet is stored in the node with the determined rank, and the determined rank is within a predetermined first rank range. In some cases, an analysis unit that sets a media packet to be referred to by a node constituting a node sequence including a node having the highest rank as an output target media packet;
When the output target media packet includes an unreceived media packet, the stored FEC packet is used to restore the unreceived media packet into an output target media packet, and the output target media packet is set to a sequence number. Output means for outputting in order;
A forward error recovery system receiver characterized by comprising: The forward error recovery method receiver according to claim 1,
When the output means outputs the output target media packets in the order of the sequence number, the analysis means invalidates the storage contents of the node storing the storage position of the output media packet and raises the rank of the subsequent node. A forward error recovery receiver. The forward error recovery method receiver according to claim 1,
When the permutation of the determined nodes is a predetermined second or higher rank, the analyzing means does not set a media packet to be referred to by a node constituting a node string including the node with the highest rank as an output target media packet. A forward error recovery system receiver characterized by moving up the order of subsequent nodes. The forward error recovery method receiver according to claim 1,
In the case where the determined rank is in a predetermined first rank range, the analyzing means refers to a node that forms a node string having a predetermined relationship with the node string in addition to the node string that includes the node having the highest rank. A forward error recovery system receiver characterized in that a media packet to be output is an output target media packet. The forward error recovery method receiver according to claim 4,
The forward error recovery method receiving apparatus according to claim 1, wherein the node sequence having the predetermined relationship is a node sequence including a node having the highest rank other than a node constituting the node sequence including the node having the highest rank. The forward error recovery method receiver according to claim 1,
When the FEC packet has not been received, the analysis means stores the storage position of the media packet received first in a predetermined node, and determines the node rank corresponding to the sequence number of the media packet after receiving the FEC packet. And a forward error recovery receiver apparatus that shifts the ranks assigned to the nodes so that the predetermined nodes have the determined node ranks. A forward error recovery method reception output method of receiving a stream of media packets with a sequence number and a stream of FEC packets generated based on the media packets and outputting the media packets in order of sequence numbers,
A buffering step of storing the received media packet and FEC packet in a buffer;
A node setting step for setting a plurality of node strings composed of a plurality of nodes ranked in correspondence with the sequence numbers in order to refer to the storage position of the media packet in the buffer;
A node storage step of determining a rank of a node corresponding to the sequence number of the received media packet, and storing a storage position of the media packet in a node to which the determined rank is attached;
When the determined rank is the predetermined first rank range, a media packet to be referred to by a node constituting a node sequence including the node with the highest rank is set as an output target media packet, and the output target media packet is not received. An output step of restoring the unreceived media packet to the output target media packet using the stored FEC packet, and outputting the output target media packet in the order of the sequence number.
A forward error recovery system reception output method comprising: The forward error recovery method reception output method according to claim 7,
When the media packets to be output are output in order of sequence number, the storage contents of the node that stores the storage position of the output media packet are invalidated, and the update step further advances the rank of the subsequent node. Forward error recovery method reception output method. The forward error recovery method reception output method according to claim 7,
If the determined permutation of the nodes is equal to or higher than the predetermined second order, the subsequent node without using the media packet to be referred to by the node constituting the node train including the node with the highest rank as the output target media packet The forward error recovery method reception output method, wherein the rank is moved up. The forward error recovery method reception output method according to claim 7,
In the output step, when the determined rank is the predetermined first rank range, in addition to the node string including the node with the highest rank, the nodes constituting the node string having a predetermined relationship with the node string are referred to A forward error recovery system reception output method, wherein a media packet to be output is a media packet to be output. The forward error recovery method reception output method according to claim 10,
The forward error recovery method reception output method, wherein the node string having the predetermined relationship is a node string including a node having the highest rank other than the nodes constituting the node string including the node having the highest rank. The forward error recovery method reception output method according to claim 7,
In the node storing step, when the FEC packet has not been received, the storage position of the media packet received first is stored in a predetermined node, and after receiving the FEC packet, the node rank corresponding to the sequence number of the media packet is set. A forward error recovery method reception output method, characterized in that the order given to the nodes is shifted so that the judgment is made and the predetermined nodes have the judged node order.