KR20130044782A - Encoding method and apparatus of forward error correction in a multimedia system - Google Patents

Abstract

PURPOSE: A FEC(Forward Error Correction) encoding method and an apparatus for the same in a multimedia system are provided to generate a parity block having various lengths. CONSTITUTION: A MMT D.1 layer(601) comprises a source block by being inputted to a MMT package from a MMT E.1 layer(603) in an encoding process and being divided into a fixed interval. An AL(Application Layer)-FEC converter(605) converts a source block into a 2D array comprised of information payloads having a same length. A FEC encoder(607) generates a parity block from an information block by performing a FEC encoding, and transmits to a payload format generator(609). The payload format generator adds the parity block to the source block, adds a payload header to each payload, and transmits a packetized MMT payload format to a MMT D.2 layer or an IETF(Internet Engineering Task Force) application protocol layer(611).

Description

ENCODED METHOD AND APPARATUS OF FORWARD ERROR CORRECTION IN A MULTIMEDIA SYSTEM}

The present invention relates to an FEC encoding method and apparatus for supporting data of various lengths in a multimedia system.

Data congestion on the network has been intensifying due to the increase in various contents and the increase in high-capacity contents such as high definition (HD) content and ultra high definition (UHD) contents.

1 and 2 are schematic diagrams illustrating a general IP-based network topology and data flow.

Referring to FIG. 1, due to such a situation, contents sent from a transmitter (Host A) 101 are not delivered to the receiver (Host B) 103 and a part of contents is lost on a route. Occurs. In general, since data data is transmitted in packet units, data loss occurs in transport packet units.

"Data" in FIG. 2 is regarded as RTP packet data after packetizing the compressed data at the AV Codec stage using, for example, Real Time Protocol (RTP) (see RFC3550, RFC3984 document of IETF) or MPEG MMT. It can be understood as MMT packet data being progressed in (MPEG Media Transfort).

Since the receiver 103 of FIG. 1 cannot receive the lost transmission packet on the network, it is impossible to know the data in the lost transmission packet, thereby degrading the quality of the audio, deteriorating the video quality or breaking the screen, missing the subtitles, and loss of the file. In various forms such as the user will cause inconvenience.

Thus, there is a need for a method for recovering lost data on the network. To this end, a parity block by forward error correction (FEC) encoding is added to a source block composed of a predetermined number of packets and transmitted.

In particular, in the case of AV data, a certain level of data rate must be maintained for continuous playback without interruption. This data rate depends on the performance of the Codec used during compression and the resolution of the AV data being serviced.

In addition, in the case of AV data (particularly real-time AV data), it is better to minimize delay from acquisition of data such as photographing to use by the user. This delay can occur at various stages of the overall system.

In particular, when applying AL (Application Layer) -FEC, when a packet is lost in the source block, a certain number of parity packets of the parity block must be received to recover the packet in the lost source block. Can be. Due to this characteristic, AL-FEC causes a delay in time. The delay time by AL-FEC is mainly determined by the data size of the source block and the parity block added for recovery (ie, the size of the FEC Delivery Block).

For this reason, it is desirable to configure the source block of the FEC delivery block according to the data rate and delay time required by the service that the service wants to use, and to transmit it by AL-FEC encoding. ) Codec is used according to various types of services such as 3D, Ultra High Definition (UHD), and the required data rate is different. Therefore, there is a need for an AL-FEC encoding method according to the size of various source blocks.

That is, a method of efficiently encoding data of various sizes is required when transmitting data in a packet unit by using a transmission protocol such as the RTP packet or the MMT packet.

The present invention provides a method and apparatus for efficiently AL-FEC encoding data of various lengths in a multimedia system.

In addition, the present invention provides a method and apparatus for efficiently encoding data of various sizes when transmitting data in a packet unit in a multimedia system.

According to an embodiment of the present invention, an FEC encoding method in a multimedia system using packet-based data transmission includes a process of constructing a source block including a plurality of source payloads, and a plurality of source blocks using one FEC code. Constructing a parity block including parity payloads of the parity payload; and configuring a FEC transport block including the source block and the parity block, wherein the number of the plurality of source payloads is K; When the number of parity payloads is P, the plurality of information symbol portions of the source block are arranged in an mxK array, the plurality of parity symbol portions of the parity block are arranged in an mx P array, and the number of given payloads K The value of m is adjusted with respect to the FEC transport block.

In addition, the FEC encoding method in a multimedia system using packet data transmission according to an embodiment of the present invention, K source using one FEC (N_m, K_m) code for packet data transmission or data storage In generating a parity block composed of P parity payloads from a source block composed of payloads, m symbol rows of each source block for a maximum integer m satisfying mx K ≤ N_m and mx P ≤ N_m-K_m A parity symbol portion is generated from each information symbol portion using the FEC (N_m, K_m) code from at least one information symbol portion composed of a), and P parity payloads are generated from the generated at least one parity symbol portion. Configures a parity block consisting of.

According to the present invention described above, it is possible to generate a parity block having various lengths by encoding source blocks having various lengths with one FEC code. For this reason, according to the present invention, it is possible to minimize the system burden by simplifying the configuration of the FEC encoder / decoder of the transmitting and receiving end by not only providing a better quality service to the user by exhibiting better FEC performance.

1 and 2 schematically show a general IP-based network topology and data flow;
3 and 4 are diagrams for explaining an encoding method using an FEC code according to an embodiment of the present invention;
5 is a diagram illustrating an example of a structure of an MMT system and a specific transmission function layer to which the present invention is applied;
6 is a functional block diagram for explaining an encoding / decoding process of an AL-FEC source block in an MMT system according to an embodiment of the present invention;
7 illustrates an information symbol part and a parity symbol part in an FEC transport block generated by AL-FEC encoding according to an embodiment of the present invention;
8 is a diagram illustrating an example of a method of generating a parity symbol part using a first interleaving method and an FEC (7680, 6400) code according to an embodiment of the present invention;
9 is a diagram illustrating an example of a method of generating a parity symbol portion using a second interleaving method and FEC (7680, 6400) codes according to an embodiment of the present invention;
10 is a diagram showing a two-stage FEC coding structure according to an embodiment of the present invention;
FIG. 11 is a diagram illustrating an example of an FEC block and an FEC cluster structure generated by the two-stage FEC encoding structure of FIG. 10; FIG.
12 and 13 are diagrams comparing the performance of the LDPC code when FEC coding is applied on a random erase channel according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

Hereinafter, an embodiment of the present invention proposes a method of efficiently AL-FEC encoding data of various sizes in the case of transmitting data in a packet unit using a transmission protocol such as an RTP packet or an MMT packet.

Prior to describing an embodiment of the present invention, terms used in the present invention will be defined as follows.

FEC packet block: a set of FEC packets for transmitting the FEC transport block

FEC packet: A packet consisting of a header and a payload for transmitting a parity payload as well as a source payload.

FEC transport block: a set consisting of a source block and a parity block

Source block: set of source payloads

Parity block: a set of parity payloads

Source payload: payload of the FEC packet for transmitting source symbols

Information payload: A payload consisting of source symbols and possible padding data

Parity payload: payload of an FEC packet for transmitting parity symbols

Symbol: The unit of data, its size (in bits) is referred to as a symbol size.

Information symbol part: A set of source symbols and possible padding data that are considered together for FEC encoding

Parity symbol portion: a set of parity symbols generated by FEC encoding from an information symbol portion

FEC frame: One codeword composed of information symbol part and parity symbol part

Table 1 below is an excerpt of a known Internet Engineering Task Force (IETF) RFC5170 document.

Number of packets G Symbol size k 4000≤nb_pkts One pkt_sz 4000≤k 1000≤nb_pkts <4000 4 pkt_sz / 4 4000≤k <16000 500≤nb_pkts <1000 8 pkt_sz / 8 4000≤k <8000 1≤nb_pkts <500 16 pkt_sz / 16 16≤k <8000

Referring to Table 1, the number of packets is divided into 1, 4, 8, 16 groups G according to the number of packets nb_pkts, and the number of packets is expanded by a multiple of the number of groups. An example of constructing a block and encoding using a randomly generated LDPC H matrix using a pseudo random number generator (PRNG) (not shown) according to the length k of each source block is shown. will be.

In general, AL-FEC encoding encodes one symbol per packet to correspond to one symbol of an FEC codeword. In the existing multi-symbol per packet scheme, a plurality of symbols per packet are included in one codeword.

The receiver may receive the seed value of the pseudo random number generator (PRNG) and the length k value of the source block from the transmitter to generate an LDPC H matrix corresponding to a software code (S / W code) to recover lost packets. .

However, since the multi-symbol per packet method can also have a plurality of k values, a plurality of LDPC H matrices are required for each k value. In addition, even if the length k value of the source block is fixed, more LDPC H matrix is required because the LDPC H matrix varies according to the addition rate.

In consideration of this, according to an embodiment of the present invention, in a system for transmitting and receiving data in packet units through a wired or wireless communication network, the number K of information payloads for an information block of data to be transmitted by applying AL-FEC is at least two or more. In the case where the number P of parity payloads for a parity block added to the information block may have a value or may have at least two values, K in one FEC (K_m + P_m, K_m) code We propose an AL-FEC coding scheme for generating the P parity block in the information block consisting of two pieces.

(K_m is the number of symbols of the information portion of the FEC (K_m + P_m, K_m) code and P_m is the number of symbols of the parity portion of the FEC (K_m + P_m, K_m) code. K_m + P_m = N_m .)

The method for generating an information block consisting of K information payloads from a source block consisting of K 'source payloads is a method according to IBG_Mode (Information Block Generation Mode) according to the applicant's prior application No. 10-2011-0104870. Is described.

Briefly describing the method according to IBG_Mode, if IBG_Mode = 0, it indicates that the length of the source payload is the same and a parity block is generated from the same information block as the source block to configure and transmit the FEC transport block. If IBG_Mode = 1, an information block is generated by the existing method to perform AL-FEC encoding. The conventional method is a method of generating a parity block after adding padding data to make each payload size equal to a payload having a maximum length when source payloads having a variable packet size are input.

If IBG_Mode = 2, a source block composed of K 'source payloads having a variable length is sequentially disposed in an information block that is a two-dimensional array consisting of K information payloads having the same length. The length block includes length information of each source payload of the source block. Further, a parity block composed of parity payloads having the same length is generated from the information block, and a length parity block is generated from the length block. Accordingly, the FEC transport block is AL-FEC encoded including a source block, a length parity block, and a parity block.

Hereinafter, a method of encoding using one FEC (N_m, K_m) code according to an embodiment of the present invention will be described.

When generating a parity block composed of P parity payloads (consisting of S_max symbols) from an information block consisting of K information payloads (consisting of S_max symbols), the information block is composed of q information symbol parts. The value of q is set to satisfy the equations qx K_m ≥ S_max x K and qx P_m ≥ S_max x P.

When dividing the information block into q information symbol parts with respect to the set q value, the sizes (number of symbols) of each information symbol part are equally equal (e.g., the deviation of the sizes is 1 or less). Each information symbol portion is composed of q information symbol portions so as to equalize the number of symbols selected from each of the K information payloads of the information block as much as possible (e.g., the deviation of the numbers is 1 or less). (Where the size of the ith information symbol part (number of symbols) is defined as K'_m (i) (≤K_m), i = 1,2, ..., q)

A parity symbol portion consisting of P′_m (i) (≦ P_m) parity symbols is generated from each of the q information symbol portions by using an FEC (N_m, K_m) code or a shortened code thereof. (Where P'_m (i) is the number of parity symbols generated from the ith information symbol part, so that the maximum equal number is obtained for i = 1, 2, ..., q, for example, parity symbols) The number of variations is less than 1)

Herein, the short code (Shorten code) may be referred to as a FEC (N_m-s, K_m-s) code, and means that the information portion of the FEC (N_m, K_m) code is shortened by s symbols. .

And mapping the generated q parity symbol portions to a parity block composed of P parity payloads, and when P'_m (i) parity symbols of any parity symbol portion are mapped to P parity payloads, Make the number of parity symbols mapped in the parity payload as even as possible (where the number of parity symbols is less than 1).

And where K'_m (1) + K'_m (2) + ... + K'_m (q) = S_max x K and P'_m (1) + P'_m (2) + ... + P It satisfies that '_m (q) = S_max x P'.

3 and 4 illustrate an encoding method using FEC (N_m, K_m) codes according to the embodiment of the present invention described above.

As described above, the information block 301 composed of K information payloads is divided into q information symbol portions 303 as shown in FIG. 3, and each information symbol portion is obtained by parity using FEC (N_m, K_m) codes. A symbol portion is generated as shown in FIG. Each generated parity symbol portion is mapped to a parity block composed of P parity payloads 305 as shown in FIG. 3.

Each information symbol portion is composed of m symbols or m + a symbols of each information payload, as shown at 303 of FIG. The parity symbol portion is also assigned to m symbols or m + a symbols of each parity payload, as shown at 305 of FIG.

As a result, lost data in the payload due to packet loss is divided as evenly as possible into each information symbol portion of an information block, thereby stabilizing FEC performance as much as possible. It is possible to decode FEC frame as stably as possible with m + a symbol loss (where a = 0 or 1).

In addition, according to the AL-FEC encoding method of the present invention, it is possible to encode a variety of values of K and P with one FEC (N_m, K_m) code, which simplifies the system structure and minimizes the burden on the transceiver. In addition, since a single FEC code is encoded, hardware (H / W) can be implemented and the implementation burden can be reduced.

The single FEC code can be implemented in both S / W and H / W implementations. In general, H / W implementations have a faster FEC encoding / decoding rate, so H / W implementations are better for high-speed data transmission such as UHD. Suitable.

Meanwhile, in the case of K_m <K in FIG. 3, two information symbol parts are alternately arranged in the information block. For example, if K_m = 3200 and K = 6400, each row of the information block consists of two information symbol parts, one information symbol part is composed of 3200 odd-numbered symbols, and the remaining information symbol part is even-numbered. It consists of 3200 symbols. In the parity symbol portion of the parity block, each symbol of two parity symbol portions is alternately mapped to a row on the parity block.

In the case of K_m = 3200 and K = 4800, two information symbol parts are generated using 2400 odd-numbered symbols and 2400 even-numbered symbols of each row of the information block, and the 800 symbols are filled with zeros during encoding. After that, shortening is performed. The generated parity symbol portion is also punctured if necessary and then allocated to the parity block.

5 illustrates an example of a structure of an MPEG Media Transport (MMT) system and a specific transport function layer to which the present invention is applied.

5 shows a structure of the MMT system, and 503 shows a detailed structure of a transport function layer of the MMT system.

Referring to FIG. 5, audio / video data compressed from a media coding layer is passed through an encapsulation function layer (hereinafter referred to as an E layer) to a file format. Packaged and output in a similar form as in the following, in the delivery function layer, the output of the E layer is converted into an MMT payload format, and then an MMT transport packet header is added to the MMT payload. Output in MMT transport packet or RTP packet using existing RTP protocol.

Thereafter, the RTP packet is transmitted through a transport protocol layer of User Datagram Protocol (UDP) / Transmission Control Protocol (TCP) and finally IP packetized in the Internet Protocol (IP) layer.

Also, referring to FIG. 5, the AL-FEC is applied to a process of generating the MMT payload format in the MMT D.1 layer by receiving the output from the E layer. When the FEC is applied, the control function layer is If FEC encoding is to be applied, this is instructed to a delivery function layer, and the MMT D.1 layer and the MMT D.2 layer generate and output an FEC packet block. And the MMT D.3 layer provides a function related to information transmission between layers. In other words, the MMT D.3 layer enables content delivery and necessary communication between layers.

FIG. 6 is a functional block diagram illustrating an encoding / decoding process of an AL-FEC source block in an MMT system according to an embodiment of the present invention. Referring to FIG. 6, in encoding, an MMT D.1 layer 601 is an MMT. E.1 Layer 603 receives MMT Package (format created for the purpose of storing AV data, file, text, etc. in storage or considering transfer) and divided into certain unit (ie, Source Payload) for transmission. Configure The AL-FEC converter 605 converts the source block into an information block that is a two-dimensional array of information payloads having the same length.

In FIG. 6, the FEC encoder 607 generates a parity block from an information block by performing FEC encoding by the encoding method according to an embodiment of the present invention and transmits the parity block to a payload format generator 609. The payload format generator 609 adds the parity block to a source block and adds a payload header (PLH) to each payload to convert the packetized MMT payload format into MMT D. .2 layer or Internet Engineering Task Force (IETF) Application Protocol Layer (611). Thereafter, the transport packet is transmitted by adding a UDP header and then adding an IP header by a transport protocol such as UDP of FIGS. 1 and 2.

In decoding in FIG. 6, the payload format generator 609 receives MMT payload formats in a packet received from the MMT D.2 layer or the IETF application protocol layer 611 and removes the payload header (PLH). Input source and parity payloads to the AL-FEC converter 605. The payload format generator 609 then informs the AL-FEC converter 605 of the location of the lost payloads.

Here, the positions of lost payloads are known by the receiver using ordering information such as a sequence number stored in a payload header (PLH) or an MMT header or an IETF application layer header during encoding, and the AL-FEC The converter 605 may be informed or from this information the AL-FEC converter 605 may know.

The AL-FEC converter 605 converts the source block of the FEC transport block having the input lost payload back into an information block, constructs an FEC block, decodes the information payload, and recovers the recovered information block. Switch to the source block and output to the payload format generator 609.

The payload format generator 609 sends the input source block to the E.1 layer to allow the user to proceed with the process for consuming data.

Tables 2 to 4 show that the number of payloads varies according to delay time, payload size, and data rate.

Payload size is constrained by the maximum transfer unit (MTU). Most Ethernet LANs use an MTU of 1500 bytes. Therefore, the actual payload size excluding various headers is limited to 1000 bytes, and also the long payload is used to reduce the amount of overhead caused by the header for high speed data transmission such as Full HD, 3D, and UHD. It is advantageous to.

In the case of MPEG4-AVC / H.264 codec, for example, 8Mbps is required for Full HD service, UHD service having 8K (4096 x 8192) resolution is possible at 128Mbps. If a high-compression codec such as High Efficiency Video Codec (HEVC) is used, it is known that up to twice the MPEG4-AVC / H.264 codec is possible. Taking this into consideration, an 8K UHD service can be serviced within a 400 ms delay with a payload of 12800 or 6400.

Table 2 below illustrates the number of payloads for a source block according to the payload size at a delay time of 100 ms.

Table 3 below illustrates the number of payloads for a source block according to the payload size at a delay time of 200 ms.

Table 4 below illustrates the number of payloads for a source block according to the payload size at a delay time of 400 ms.

Equation 1 below to calculate the payload number in Tables 2 to 4.

&Quot; (1) &quot;

K = R x D / S, where R = transmit bits per second (bps),

D = delay time in seconds, S = Payload Size in bits.

As a result, the maximum number of payloads required in the system may be determined according to delay time, payload size, and data rate.

Hereinafter, an embodiment of the present invention describes an encoding method when the payload size is Smax, the maximum payload is 6400, and the maximum number of parity payloads to be added is 1280.

-FEC code: (7680, 6400) code (Various codes such as LDPC, Raptor, Raptor Q can be applied.)

Information block length: K ≤ 6400

Parity block length: P ≤ 1280

-Payload Size: Given S (in bytes) (where 8 * S is a multiple of m), find the maximum integer value m that satisfies m x K ≤ 6400 and m x P ≤ 1280.

In this case, the information symbol portion and the parity symbol portion in the FEC transmission block generated by AL-FEC encoding according to the embodiment of the present invention are composed of each m bit rows of the information block and the parity block as shown in FIG. 7.

Hereinafter, a method of generating a parity symbol portion from each information symbol portion using, for example, FEC 7880 and 6400 codes will be described.

8 illustrates an example of a method of generating a parity symbol part by using a first interleaving method and an FEC (7680, 6400) code according to an embodiment of the present invention.

Referring to FIG. 8, in the information symbol portion of each mx K array, ordered mx K bits by first numbering K bits of the first row from left to right and numbering K bits of the second row from left to right. After interleaving the information symbol part composed of 6400-(mx K) bits with 0 and configuring the information part of 6400 bits, the parity part 1280 bits are generated by FEC (7680,6400) code, and then 1280-(mx P) Puncturing the bits to generate a parity symbol portion consisting of mx P bits. Again, this is the parity symbol portion of the mx P array on the parity block. Shows mapping.

9 illustrates an example of a method of generating a parity symbol part using a second interleaving method and an FEC (7680, 6400) code according to an embodiment of the present invention.

9, from top to bottom of the first column of the information symbol portion consisting of each mx K array on the information block, from top to bottom of the second column, ..., above the K-th column Except for the method of generating the information symbol part consisting of mx K bits by numbering from the bottom to the same, and the method of mapping from the parity symbol part consisting of mx P bits to the mx P array parity symbol part corresponding to the parity block in the same manner. The method of FIG. 9 is the same as the method of FIG. 8.

By this method it is possible to FEC encode using one FEC (7680,6400) code for all Ks less than 6400 and all Ps less than 1280.

The above method is equally applicable to a two stage FEC coding structure as shown in FIG. 10.

10 is a diagram illustrating a two-stage FEC coding structure according to an embodiment of the present invention.

Referring to FIG. 10, for a two-stage FEC encoding structure, the source block 1001 is composed of M sub-blocks 1003, and each sub-block is converted into an information block, and then, a first block. Encoded by the FEC encoder 1005, a first parity block (Parity 1 Block) is added to the corresponding information block. On the other hand, the source block 1001 is converted into an information block and then encoded by the second FEC encoder 1005 to generate a second parity block. The FEC block and the FEC cluster structure generated by the two-stage FEC coding structure of FIG. 10 are the same as in FIG. This is possible by using the second FEC encoders 1005 and 1007 as the same FEC code.

E.g,

Source block length: K ≤ 6400

Sub-Block Length: K (i) ≤ 6400 (i = 1,2, ..., M)

ith Parity 1 Block Length:

  P1 (i) ≤ 1280 (i = 1,2, ..., M)

Parity 2 Block Length: P2 ≤ 1280

First, the maximum value m (i) satisfying m (i) x K (i) ≤ 6400 and m (i) x P1 (i) ≤ 1280 is found for each subblock. The first parity block is generated by using the FEC (7680, 1280) codes by the above-described FEC encoding method of the present invention using the value of m (i). replaced by m (i)

For the source block, the maximum value satisfying m x K ≤ 6400 and m x P2 ≤ 1280 is found, and the second parity block is generated using the FEC (7680, 1280) code by the FEC encoding method of the present invention.

Adding an embodiment of specific values of K, K (i), P1 (i), and P2 is as follows.

K = 3200 and P2 = 160

K (i) = 200 (i = 1,2, ..., 16)

-P (i) = 30 (i = 1,2, ..., 16)

M (i) = 32 for each subblock and m = 2 for the source block, and coded using the FEC (7680,6400) codes in the same manner as in FIGS. 7, 8, and 9 above.

12 and 13 compare the performance of the LDPC code when FEC encoding is applied on a random erase channel according to an embodiment of the present invention. In detail, FIG. 12 shows, for example, K = 200 and FIG. 13 compares the performances of the FEC (7680,1280) codes, that is, the LDPC (7680, 6400) codes, according to an embodiment of the present invention. will be.

Assuming that K = 200 or 400 and P / K = 5,10,15,20%, LDPC6400_5%, LDPC6400_10%, LDPC6400_15%, and LDPC6400_20% in Figs. 12 and 13 are FECs according to an embodiment of the present invention. The coding scheme is applied when K = 200 and K = 400. LDPC200 means a code having an information length of 200, and LDPC400 means a code having an information length of 400. LDPC200 and LDPC400 correspond to the case where m = 1 in the embodiment of the present invention, LDPC6400 corresponds to the case where m = 32 for K = 200, and m = 16 for K = 400.

12 and 13, it can be seen that the performance of the FEC coding scheme according to the embodiment of the present invention is better.

When the maximum value of K is K_m and the maximum value P_m of P, at least a FEC (K_m + P_m, K_m) code is required from a transceiver perspective.

Therefore, by applying the embodiment of the present invention, it is possible to apply coding to various K values and P values with one FEC (K_m + P_m, K_m) code, so that the system can be simplified and provide better performance. It can provide high quality contents to users.

Table 5 below shows the MMT payload format.

Payload Header MMT Payload (Source Payload or Parity Payload)

Table 6 below shows the payload header format.

 Payload Type  Sequence Number  FEC_Flag  Block Boundary Info  FEC_Type_Info  Interleaving Mode  FEC Delivery Block Length: N  Source Block Length: K

In Table 6, "Payload Type" indicates whether the payload of the corresponding MMT payload format is a source payload or a parity payload.

"Sequence Number" is sequentially increased or decreased in order to indicate the order of payloads to be transmitted, and it is possible to know whether a packet is lost from the Sequence Number.

"FEC Flag" indicates whether FEC is applied. For example, if the value is 0, only the source block is transmitted without the parity payload, and if the value is 1, the parity block is added to the source block, that is, the FEC scheme is applied and transmitted. When the two-stage FEC coding structure is applied, it is preferable to extend the FEC flag field to distinguish one stage or two stages.

"Block Boundary Info" informs the boundary of the FEC transport block, and assigns the sequence number value of the first source payload of the FEC transport block to all headers.

"FEC_Type_Info" indicates what the applied FEC code is. 0 can be assigned to LDPC, 1 to Raptor, and 2 to RaptorQ. If one FEC code according to an embodiment of the present invention is not predetermined in the transmission / reception period, the code length and information length (or information length and parity length) of the FEC code are stored in this field and transmitted.

"Interleaving Mode" indicates an interleaving mode applied by an embodiment of the present invention. For example, 0 indicates interleaving in FIG. 8, and 1 indicates interleaving in FIG. 9.

"FEC Delivery Block Length" refers to the number of payloads included in the FEC transport block.

The source block length refers to the number of source payloads included in the source block.

In the interleaving mode determination, if the FEC code of the quasi-cyclic LDPC (QC-LDPC) structure is used, the method of FIG. 8 is more preferable than the method of FIG. 9, and the FEC code of the random LDPC structure is the method of FIG. More preferred.

For example, the reason why the scheme of FIG. 8 is preferable in the QC-LDPC structure is, for example, K = 200, K_m = 6400, and the mother matrix of the QC-LDPC (6400 + P_m, 6400) code is LDPC (200 + P, 200) The subblock size of the QC-LDPC code is composed of a 32 x 32 matrix. When the coding of the present invention is applied to the QC-LDPC (6400 + P_m, 6400) code using the interleaving method of FIG. 9, m = 32, and 32 bits are lost when Payload is lost, which is a subblock 32 on the QC-LDPC H matrix. x 32 is erased.

Since this corresponds to 1 bit erasure on the mother matrix of the QC-LDPC code, the performance is the same as that of the mother matrix. In the case of the random LDPC structure, the interleaving scheme as shown in FIG. 9 is more suitable because the scheme of FIG. 9 is simpler in shortening / puncturing.

At least one of the information of the payload header format shown in Table 6 is necessary information for AL-FEC and may be omitted if there is information overlapping with the header of the MMT D.2 layer or the application layer header. That is, the information required for AL-FEC is not limited to being stored in the payload header format, but may be stored in the header of the MMT D.2 layer or an application layer header such as RTP, FLUTE, and HTTP if necessary.

In particular, since the RTP header includes the "Sequence Number" and "Payload Type" fields, it is possible to minimize the overhead by utilizing this. In this case, the "Payload Type" information of the RTP header makes it possible to distinguish between the source payload and the parity payload, and the "Sequence Number" is set to include the role as "Sequence Number" in Table 6 above. "Block Boundary Info" of 6> only needs to set the sequence number of the header of the RTP packet including the first payload of the FEC transport block. In addition to RTP, if the MMT header includes sequence number or payload type information, the same applies. Table 7 below shows a protocol stack transmitted using an IP protocol as an example of a payload header format for parity payload.

 IP Protocol Header  UDP / TCP Header  MMT Protocol / Application Protocol (e.g. RTP)  MMT Payload Format Header  MMT Payload (Source Payload / Parity Payload)

Claims (2)

In the FEC encoding method in a multimedia system using packet data transmission,
Constructing a source block comprising a plurality of source payloads;
Constructing a parity block including a plurality of parity payloads from the source block using one FEC code; And
Configuring a FEC transport block including the source block and the parity block;
When the number of the plurality of source payloads is K, and the number of the plurality of parity payloads is P, a plurality of information symbol portions of the source block are arranged in an mxK array, and a plurality of parity symbol portions of the parity block are provided. Is arranged in an mx P array, and adjusts the m value for a given number K of payloads to configure the FEC transport block.
The method of claim 1,
Each information symbol portion of the source block is composed of m number of symbol rows, and when the FEC code is called FEC (N_m, K_m) code,
M is a maximum integer satisfying mx K ≤ N_m and mx P ≤ N_m-K_m,
And generating a parity symbol part from each information symbol part by using the FEC (N_m, K_m) code to configure the parity block.

Images