KR102021872B1 - Raptor-q encoding apparatus with improved encoding delay time and method thereof - Google Patents

Abstract

According to the present specification, disclosed are an encoding apparatus and method which can improve raptor-Q encoding delay time for a source block generated in real time. According to the present specification, the encoding apparatus and method perform an operation for generating an intermediate symbol using only source symbols generated up to now, even before all source symbols constituting the source block are generated. Then, operations can be performed in advance as additional source symbols are generated, and about 90% of the operations can be performed in advance just before all the source symbols constituting the source block are generated.

Description

RAPTOR-Q ENCODING APPARATUS WITH IMPROVED ENCODING DELAY TIME AND METHOD THEREOF

The present invention relates to an apparatus and method for Raptor Q encoding, and more particularly, to an apparatus and method for Raptor Q encoding with improved recovery symbol generation delay time in real-time generated data.

Unlike the Transmission Control Protocol (TCP) transmission method, the UDP transmission method does not guarantee stable transmission of data packets.

Conventionally, in this case, the receiver uses the ARQ (Automatic Repeat reQuest) method to identify a particular packet that is lost and request a retransmission thereof to the transmitter, and then use the ARQ method. There is this.

Recently, the transmitting device generates and transmits data containing recovery data that can recover even if the original data is lost during transmission.If the receiving device side receives some loss, the received data including the recovery data is received. It uses AL-FEC (Application Layer Forward Error Correction) encoding technology that can restore the original data from the data.

Among the AL-FEC coding techniques, there is a technique having a structural code (systematic code) characteristics.

An AL-FEC encoding technique that does not have structural code characteristics is, for example, LDPC AL-FEC technique of RFC 5170 "Low Density Parity Check (LDPC) Staircase and Triangle Forward Error Correction (FEC) Schemes". For reference, assuming that 10% of recovery data is required for every 100 raw data, A1, A2,... , Based on the raw data from A100, a total of 110 totally new data B1, B2,… , B110 is generated and transmitted to the receiving apparatus. If the receiving device succeeds in recovering the original data from the received data, A1, A2,... , A100 recovers all but fails to recover the original data. However, neither A100 can recover.

The AL-FEC coding technique having a structural code characteristic is, for example, a RaptorQ AL-FEC technique, which includes A1, A2,... Based on the raw data of A100, a total of 110 pieces of data are transmitted, 100 of which are A1, A2,... , The original data of A100 is used as it is, and B101, B102,... , Only B110 is generated so that A1, A2,... , A100, B101, B102,... , 110 data of B110 are transmitted to receiver. Where A1, A2,... , A100 are source symbols, B101, B102,... , B110 is a recovery symbol. The receiver receives less than 100 of the total 110 data, and A1, A2,... Even if the original data of A100 cannot be restored, A1, A2,... The data received at A100 can be used effectively as it is. Also, the receiver receives A1, A2,... When receiving at least 100 of all 110 data without distinguishing the source symbol and the recovery symbol. The entire original data can be restored even if data not received during the A100 occurs. However, even if the receiving device receives more than 100 pieces of data, there is a probability that the data restoration fails. For example, if 100 are received, the probability of restoration is 1/100, and if 101 are received, the probability of failure is significantly reduced to 1/10000.

The AL-FEC coding technique having a structural code characteristic can secure a target restoration reliability by setting the amount of recovered data in consideration of the average network loss rate and the allowable restoration failure rate.

An erase channel treats all data packets as lost if the received data packets are incomplete due to an error. For example, if a UDP packet arrives in the wireless network, the entire packet is discarded if there is a CRC error due to a CRC check.

On the other hand, if a packet goes beyond the router's throughput while the packet is being sent along with packets originating from elsewhere, the router discards the packet. For UDP packets that do not undergo retransmission, packet loss may also occur due to this reason.

However, in order to apply AL-FEC, all source blocks must be prepared. Therefore, in case of transmitting live broadcast through broadcasting equipment, in order to apply AL-FEC, generation of the Raptor Q recovery symbol has to be delayed until the amount of data generated in real time becomes the amount of data corresponding to the source block. have.

Patent Publication No. 10-2017-0030235, 2017.03.17

RFC 6330 "RaptorQ Forward Error Correction Scheme for Object Delivery" RFC 5170 "Low Density Parity Check (LDPC) Staircase and Triangle Forward Error Correction (FEC) Schemes"

It is an object of the present specification to provide an encoding apparatus and a method capable of improving a raptor Q recovery symbol generation delay time for a source block generated in real time.

The present specification is not limited to the above-mentioned task, and other tasks not mentioned will be clearly understood by those skilled in the art from the following description.

According to an aspect of the present invention, there is provided an encoding apparatus, including: a padding unit configured to add an padding symbol to a source block to generate an extended source block; An intermediate symbol generator configured to generate an intermediate symbol using the extended source block; And an LT encoder for generating a repair symbol using the intermediate symbol, wherein the intermediate symbol generator comprises: a memory unit for storing an operation list according to a Gaussian elimination method; And an operation unit configured to perform an operation according to the operation list using source symbols generated up to an operation start time.

The operation list may include operation order information, first symbol information, second symbol information, and operation content information.

According to an embodiment of the present specification, the operation unit may store the operation target list managing the operation order information for which the operation is completed or the operation order information for which the operation is not completed, in the memory unit.

According to one embodiment of the present specification, the operation unit may store an unusable symbol list for managing information of a first symbol and a second symbol that are not available for operation in the operation list.

In this case, the operation unit may perform the remaining operations except the operation including the first symbol or the second symbol included in the unavailable symbol list.

In this case, if the first symbol is included in the unavailable symbol list and the operation is excluded, the second symbol of the operation may be added to the unavailable symbol list and the following operation may be continued.

The operation unit may initialize the unavailable symbol list after performing all of the operations that can be calculated in the operation list using source symbols generated up to a start point of an operation.

According to one embodiment of the present specification, the operation unit may perform an operation according to the operation list after a predetermined number of source symbols predetermined by a user is generated.

The encoding method according to the present specification for solving the above problems, the padding step of generating an extended source block by adding a padding symbol to the source block; Generating an intermediate symbol using the extended source block; And an LT encoding step of generating a recovery symbol using the intermediate symbol, wherein the intermediate symbol generation step comprises: applying a Gaussian elimination method stored in a memory unit using source symbols generated up to an operation start time; The operation step of performing the operation according to the operation list according to; may further include.

The operation list may include operation order information, first symbol information, second symbol information, and operation content information.

According to one embodiment of the present specification, the operation step may further include storing the operation target list managing the operation order information for which the operation is completed or the operation order information for which the operation is not completed, in the memory unit. have.

According to one embodiment of the present specification, the operation step may store, in the memory unit, an unusable symbol list that manages information of a first symbol and a second symbol that are not available for calculation in the calculation list.

In this case, the operation may perform the remaining operation except for an operation including the first symbol or the second symbol included in the unavailable symbol list.

In this case, if the first symbol is included in the unavailable symbol list and the operation is excluded, the second symbol of the operation may be added to the unavailable symbol list and the subsequent operation may be continued. The operation step may initialize the unavailable symbol list after performing all of the operations that are operable in the operation list by using the source symbols generated up to an operation start time.

According to an embodiment of the present disclosure, the operation may be performed according to the operation list after the number of preset source symbols is generated.

Other specific details of the invention are included in the detailed description and drawings.

According to the present specification, when a source block is generated in real time, an encoding delay time may be improved as compared with a conventional Raptor Q encoding scheme.

Effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a simplified block diagram of a Raptor Q encoder and a Raptor Q decoder.
2 is a reference diagram of the A matrix.
3 is a block diagram of an intermediate symbol generator according to the present specification.
4 is an exemplary diagram of an operation list according to the present specification.
5 is a reference example of a symbol to be excluded.
6 is a reference diagram of an operation rate according to the number of generated source symbols.
7 is a reference diagram for the efficiency of the encoding apparatus according to the present specification.

Advantages and features of the invention disclosed herein, and methods of achieving them will be apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present disclosure is not limited to the embodiments disclosed below, but may be implemented in various forms, and the present embodiments are merely provided to make the disclosure of the present disclosure complete, and those of ordinary skill in the art to which the present disclosure belongs. It is provided to fully inform the skilled person (hereinafter, "the person in charge") the scope of the present specification, the scope of the present specification is defined only by the scope of the claims. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The encoding method (hereinafter, referred to as 'encoding method') for AL-FEC according to the present specification follows the Raptor Q code described in "RFC 6330-RaptorQ Forward Error Correction Scheme for Object Delivery." Thus, this specification refers to all contents set forth in the RFC 6330 document and may be incorporated.

In order to help understand the encoding method according to the present specification, the raptor Q code will be briefly described. However, the part described in this specification is for understanding the encoding method according to the present specification, and will be simply described as necessary to understand the encoding method according to the present specification. Definitions of terms used in the present specification are as follows.

Source block: Data to be encoded / decoded according to the Raptor Q method as data to be transmitted. Can be divided into K source symbols.

Extended Source block: K '(K' ≥ K) source symbols, consisting of a source block and a padding symbol.

Symbol: a unit of data. Symbol types include source symbols, padding symbols, recovery symbols, and intermediate symbols, all of the same size.

Source symbol: The smallest unit of data used in the encoding process. Corresponds to information in the source block.

Padding symbol: A symbol added to make a source block an extended source block. All data values are filled with '0'.

Repair symbol: A symbol generated by the source block. Symbol for recovering source blocks in anticipation of loss during transmission.

Intermediate symbol: A symbol generated from a source symbol, which is generated in an intermediate process for generating a repair symbol. In the Raptor Q decoder, it is also a symbol that is generated in the middle of generating a lost source symbol.

1 is a simplified block diagram of a Raptor Q encoder and a Raptor Q decoder. Referring to FIG. 1, an encoding process and a decoding process of a Raptor Q code will be briefly described.

The transmitter includes a Raptor Q organizational encoder (hereinafter referred to as Raptor Q encoder) 100 that encodes the source block M using the Raptor Q code. In detail, the raptor Q encoder 100 may include a padding unit 110, an intermediate symbol generator 120, and a LT (Luby transform) encoder 130.

The padding unit 110 adds a padding symbol (padding data) to the source block M to generate an extended source block M '. Encoding according to the Raptor Q code requires a predetermined number of source symbols (see 5.6 of RFC 6330). Assuming a situation in which video is encoded at a constant bit rate (CBR) and transmitted over the Internet, a source block having a specific size is generated every predetermined time, and in case of an Ethernet network, for example, the maximum size of an Ethernet packet (MTU) ) Should be sent in sizes of 1500 bytes or less. Assume that the number of source symbols generated when the source block is divided into source symbols of a certain size is K. The K source symbols may not be a predetermined number K ', K'≥K in the Raptor Q code. At this time, padding symbols filled with '0' as many as the number of short source symbols are added. As an example, it is assumed that 100 source symbols may be generated by dividing the source block M by the size of a predetermined source symbol (K = 100). According to 5.6 Table 2 of RFC 6330, the minimum number greater than or equal to 100 is' 101 '(K' = 101). Therefore, one padding symbol is added to form an extended source block M 'having a total of 101 source symbols.

The intermediate symbol generator 120 first adds S + H symbols filled with '0' in front of the extended source block M 'to form an L-dimensional column vector D, and encodes the Raptor Q code to the D vector. The inverse of matrix A is then applied (multiplied) to generate an intermediate symbol (C). Here, L is a natural number of 2 or more, and L = K '+ S + H.

2 is a reference diagram of the A matrix.

Referring to FIG. 2, the matrix A is an L × L matrix. Specifically, the matrix A is a low-density parity-check code (LDPC) matrix (G_LDPC1, G_LDPC2) having S rows, an identity matrix (I _S ) of S x S, and a high-density parity (HDPC) having H rows. -check code) comprises a matrix (G_HDPC), an identity matrix of _H x H (I H), and K 'pieces (K≤K' encoded with a line of the) matrix (G_ENC). S and H are also predetermined values by 5.6 of RFC 6330. For example, S corresponding to 101 source symbols K 'is 17 and H is 10. Thus, the A matrix can be a 128 x 128 matrix. In one example, the intermediate symbol generator 120 generates 128 intermediate symbols.

The LT encoder 130 applies the symbol generating function ENC [·] to the intermediate symbol C to generate the required number of recovery symbols (RFC 6330 5.3.5.3), and adds them with the source symbols to encode them. Generated source block, that is, message E (M). Treating source symbols and recovery symbols the same is a characteristic of Raptor Q.

The message E (M) is transmitted through an erasure channel (EC) channel, where data loss may occur.

The receiver includes a Raptor Q organizational decoder (hereinafter 'Raptor Q Decoder') 200 that decodes data (eg, received message E '(M)) using a Raptor Q code. In detail, the Raptor Q decoder 200 may restore the source block M through decoding, and may include a padding unit 210, an intermediate symbol generator 220, and an LT decoder 230.

The padding unit 210 generates a message E '(M)' by adding a padding symbol to the received message E '(M). The intermediate symbol generator 220 generates an intermediate symbol C by applying (multiplying) the inverse of the matrix A '(the Raptor Q code matrix) for decoding the Raptor Q code to the message E' (M) '. Here, matrix A 'is a = L x L matrix. However, the component value of the matrix A 'may vary depending on the number of the source symbol and the recovery symbol actually arriving at the decoder. If some of the source symbols are missing and part of the column vector D is replaced with a repair symbol, the matrix A 'is different from the matrix A used in the encoder. Specifically, the matrix A 'is an LDPC matrix (G_LDPC1, G_LDPC2) having S rows, an identity matrix (IS) of S x S, an HDPC matrix (G_HDPC) having H rows, and an identity matrix (IH) of H x H. And an encoding matrix G_ENC 'having at least K' rows.

The LT decoder 230 restores the original source block by applying a symbol generation function ENC [·] to the intermediate symbol C to generate a missing source symbol (RFC 6330 5.3.5.3).

Since general operations of the Raptor Q encoder 100 and the Raptor Q decoder 200 are already well known, a detailed description thereof will be omitted.

Returning to the encoding process, the Raptor Q encoder 100 must generate the intermediate symbol C to encode the source block. The relationship between the matrix A and the intermediate symbol (C) is as follows.

AC = D

In this case, D is a column vector having S + H symbols filled with '0's before the extended source block, that is, an L-dimensional column vector composed of source symbols. In this case, an inverse of A is required to obtain the intermediate symbol C.

C = A ^-1 D

 The problem is that it is not easy to actually calculate and compute the inverse of A. By the previous example (K '= 101), the A matrix is a 128 x 128 matrix. Furthermore, since the Raptor Q code can have up to 56,403 extended source blocks, the A matrix is 57,326 x 57,326 matrices, and there is a problem in that the inverse of the A matrix is obtained. Fortunately, there is a Gaussian elimination method to find the inverse of A. Since Gaussian elimination itself is a well-known mathematical algorithm, the principles of the algorithm will not be described herein.

Referring to 5.4.2 of RFC 6330, a method of generating the intermediate symbol C is presented. This five-step process is a process of calculating the result of the operation in which the inverse matrix of A is applied to the column vector D. The process is generally performed by the Gaussian elimination principle, but is composed of stepwise operations using the characteristics of the A matrix shape. Among these, one stage occupies more than 50% of the number of operations in five stages of operation, and more than 85% of the computation amount of a processor is consumed. The operation list of the present specification includes only one step operation.

On the other hand, there are important prerequisites in the above-described encoding process. It is assumed that the source block, that is, data to be transmitted, is input to the Raptor Q encoder 100 in a state where all of the data to be transmitted are determined. Therefore, in the case of live broadcast image data, source blocks are generated in real time, and are not input to the Raptor Q encoder 100 until the number of the number of the source blocks is fixed, so that a work for generating a recovery symbol cannot be started. The delay over time for live video data generation is an inherent delay that cannot be overcome, however fast the hardware computing power can be. Accordingly, there is a need for a more efficient encoding method in an environment in which source blocks are generated in real time.

The encoding method according to the present specification for solving the above-described encoding delay problem is to perform an operation for generating an intermediate symbol in advance using only source symbols generated so far in real time without waiting until all source blocks are generated.

3 is a block diagram of an intermediate symbol generator according to the present specification.

Referring to FIG. 3, the intermediate symbol generator 120 according to the present specification may include a memory 121 and a calculator 122.

The memory unit 121 may store an operation list according to a Gaussian elimination method.

4 is an exemplary diagram of an operation list according to the present specification.

Referring to FIG. 4, the operation list according to the present specification may include operation order information, first symbol information, second symbol information, and operation content information. The operation list includes only one step information of operation information for calculating an intermediate symbol from a source symbol according to a Gaussian elimination method. That is, when all the operations between the source symbols are completed according to the operation order described in the operation list and subsequent two to five operations, the intermediate symbol is calculated from the source symbols.

The S, H, and K 'values are previously agreed between the Raptor Q encoder and the decoder, and the specific shape of the A matrix to be used in the Raptor Q encoder is determined according to these values. The list of operations is fixed from the specific shape of this matrix A by the principles set out in RFC 6330 5.4.2.

The operation order information indicates information on the order in which the operation should be performed. The first symbol information and the second symbol information represent information of a symbol on which an operation is performed. The arithmetic content information indicates arithmetic content performed between the first symbol and the second symbol. More specifically, GF (256), that is, an operation on a Galois field of 256 elements (RFC 6330 5.7), and when the operation content information number is 0, bitwise XOR (exclusive OR) operation of the first symbol and the second symbol is performed. If the operation content information number is not 0, first, multiplication operation is performed on the first symbol and operation information information number on the GF (256), and then the bitwise XOR operation on the result and the second symbol (RFC 6330 5.7.2) is performed. .

For example, the first operation in the operation list shown in FIG. 4 is "bitwise XOR operation of the value of the first symbol of 481 and the second symbol of 36 to store the result value in the second symbol of 36". Becomes The first symbol and the second symbol both indicate a symbol that is a component of the D vector. In the above example, the first symbol 481 is symbol 481 of the D vector, and the second symbol 36 is symbol 36 of the D vector. Means.

In addition, the ninth operation in the operation list shown in FIG. 4 octect product of the octet value of the first symbol of No. 481 and the number 240 in the GF (256) at GF (256). 2 bit and bitwise XOR operation and store the result in the second symbol of 59. "

The operation unit 122 may perform an operation according to the operation list using source symbols generated up to an operation start time.

In particular, the encoding apparatus 100 according to the present specification, the intermediate symbol generator 120 does not wait until all the source symbols constituting the source block are input, and precomputes only a portion that can be pre-calculated using only the source symbols input so far. To do. Therefore, in the present specification, 'source symbols generated up to the start of operation' refers to source symbols currently stored in a memory.

As an example, it is assumed that a source symbol is generated until a certain operation start time, and there are 500 operable symbols out of a total of 1,071 symbols (when K '= 1,002). There are symbols filled with S + H '0's at the beginning of the D vector and K'-K padding symbols at the back.However, for convenience of explanation, the number of 500 symbols that can be calculated is Corresponds to symbol 0 to symbol 499. Referring to the operation list shown in FIG. 4, since the information of the first symbol and the second symbol used is all 499 or less, operation 5 may be performed until operation 5. However, since the numbers of the second symbols required for operations 6 to 8 are '650', '847' and '982', the operation is not yet possible.

On the other hand, whether the symbol is generated so far and can be used in the operation is managed as a list of unavailable symbols. The unavailable symbol list is initialized with source symbols that have not yet been generated, but source symbols that have already been generated may be added to the unavailable list during operation.

Even if the source symbol is already generated at the time of operation, if the first symbol is included in the unavailable symbol list in each operation of the operation list, the second symbol is added to the unavailable symbol list, and the source symbol not generated in the subsequent operation list and Are treated together. On the other hand, the S + H symbols at the beginning of the D vector are filled with '0', so when initializing the unavailable symbol list, they are treated like the generated source symbols, and the K'-K padding symbols after the D vector are also the same. Are treated.

Meanwhile, while the operation unit 122 performs an operation on only a part of the operation list, additionally generated source symbols may be stored in the memory. Accordingly, the operation unit 122 initializes the unusable symbol list with source symbols not yet generated until the start of a new operation, and the second symbol is already generated in each operation during the operation. Even if it is not included, if the first symbol is included in the unavailable symbol list in the operation, the second symbol is added to the unavailable symbol list, and only the operable operation is performed until the operation end of the operation list, and still generated until that point. The unusable symbol list may be initialized with the unsourced symbols, and the list may be returned to the top of the operation list to repeat the next operation. At this time, it is possible to manage whether each operation in the operation list is performed in a flag format so that the operation already performed in the previous round is not performed again. The operation unit 122 may repeat the process several times until all source symbols constituting the source block are generated, that is, until all operations in the operation list are completed.

At this time, the operation unit 122 skips the portion calculated at the start of the previous operation and performs only the portion that has not been calculated. To this end, the operation unit 122 may store, in the memory unit 121, an operation target list for managing operation order information for which the operation is completed or operation order information for which the operation is not completed. It is not necessary to re-determine whether the operation order information in which the operation is completed in the operation list is possible at the next operation time point. The operation unit 122 may make a list of the operation order information on which the operation is completed, so that the operation may be immediately skipped at the next operation time. On the contrary, the operation unit 122 may make a list of the operation order information for which the operation is not completed and move to the operation order immediately at the next operation time to perform the operation.

Meanwhile, there may be further matters to be considered in the calculation of the operation list based on the source symbols generated by the operation unit 122 up to an operation start time. A second symbol associated with an operation that cannot be performed at the start of any operation must be excluded from all subsequent operations. This will be described with reference to FIG. 5.

5 is a reference example of a source symbol to be excluded.

Referring to FIG. 5, it is assumed that the number of source symbols generated up to the current operation start time point is 500, as in the previous embodiment. Referring to the operation sequence 123, "operate the first symbol of 560 and the second symbol of 27". In the current operation time point, the operation sequence 123 cannot be operated because the first symbol 560 has not been generated yet. After that, referring to the operation order 188, "operate the first symbol 262 and the second symbol 27". At the start of the current operation, operation order 188 itself can be considered that operation is possible at the present time because symbols 252 and 27 have already been generated. However, operation order 188 should not be calculated. The operation list is the operation contents according to the Gaussian elimination method. If operation number 188 is calculated first and the result value is stored in symbol 27, then symbol 560 is generated, and then number 123 is calculated in the next iteration operation, symbol 27 should be used in operation 123. The operation 123 is executed by using a value different from. As a result, the final result for applying the inverse of the matrix A will not be calculated, and a completely different result will be calculated. Therefore, a first symbol or a second symbol related to an operation that cannot be performed at the start of any operation should be excluded from all operations afterwards, but after the first symbol appears once in a list of operations, it reappears in the operation list. Only the second symbol need be excluded by the characteristic of one-step operation not performed.

To this end, the operation unit 122 may store, in the memory unit 121, an unusable symbol list for managing information of the first symbol and the second symbol that cannot be used in the operation list. The operation unit 122 may perform the remaining operations except the operation including the first symbol or the second symbol included in the unavailable symbol list. In this case, if the first symbol is included in the unavailable symbol list and the corresponding operation is excluded, the second symbol of the operation is also added to the unavailable symbol list and the remaining operations are continued. On the contrary, the second symbol is the unavailable symbol. If it is included in the list and the operation is excluded, the first symbol of the operation is not added to the unavailable symbol list. The operation unit 122 may initialize the unavailable symbol list after performing all of the operations that are operable in the operation list by using the source symbols generated until an operation start time. The unavailable symbol list only means that an operation is impossible in performing an operation according to an operation list using only source symbols generated to date. Therefore, whether the first symbol or the second symbol is not operable or impossible by the newly generated source symbols must be newly determined and managed at the next operation start time. Therefore, the initialization of the unavailable symbol list is an operation of erasing information on symbols included in the list and identifying a component of a D vector corresponding to a source symbol not generated until that time. More specifically, since the front S + H of the D vector is a symbol filled with '0', it is always known, and the rear K'-K is also the same because it is a padding symbol filled with '0'. This means that the unusable symbol list is initialized with a number corresponding to a source symbol not yet generated among the source symbols.

On the other hand, if the number of source symbols generated up to the start of any operation is small, there may not be many operations that can be substantially operated in the operation list. For example, suppose that a component of a D vector consisting of source symbols has 1,071 dimensions. The first S + H (59 + 10) of the D vector is an O-filled symbol, the latter K'-K is also a zero-padded symbol, and since only the middle part is a source symbol, all 1,071 symbols forming the D vector are sequentially Although not generated, hereinafter, for convenience of explanation, all 1,071 will be described as source symbols sequentially generated.

6 is a reference diagram of an operation rate according to the number of generated source symbols.

Referring to FIG. 6, a box indicated by a dotted line means a number of a source symbol used for calculation, and a box and a dotted line indicate a calculated number and ratio. Referring to FIG. 4, a total of three operations are possible at the time of generating the source symbol 76, which is 0.02% of the total operations. Next, a total of four operations are possible when the source symbol 117 is generated, which is 0.02% of the total operations. Next, when the source symbol 161 is generated, a total of five operations are possible, which is 0.03% of the total operations. Next, when the source symbol 169 is generated, a total of six operations are possible, which is 0.04% of the total operations. Next, when the source symbol 171 is generated, a total of seven operations are possible, which is 0.04% of the total operations. In the case of 1,071 source symbols from 0 to 1,070, the total number of operations is 16,819. Therefore, when the number of generated source symbols is small, there will be more objects that cannot be computed than those that can be computed. In this case, the operation unit 122 may consume more resources for managing the operation target list and / or the unavailable symbol list.

Therefore, the operation unit 122 may perform an operation according to the operation list after generating a predetermined number of source symbols predetermined by a user. The preset number of source symbols may be experimentally calculated in consideration of the total amount of calculation and the amount of calculation necessary for the operation unit 122 to manage the calculation target list and / or the unavailable symbol list.

7 is a reference diagram for the efficiency of the encoding apparatus according to the present specification.

Referring to the left side of FIG. 7, when 1,070 source symbols are generated, only one last source symbol has not been generated yet. Up to this point, 15,304 operations out of a total of 16,819 operations are available, which is 90.99% of the total. Therefore, according to the prior art, it is possible to prevent a time delay of 90% compared to performing the operation after all the source blocks are generated (one of the five steps according to 5.4.2 of RFC 6330). Throughout the steps there is a significant effect of preventing 77% (90% x 85%) time delay.

Meanwhile, although the above-described encoding apparatus 100 is illustrated in a separate configuration according to each performance result to be distinguished, a processor, an application-specific integrated circuit (ASIC), another chipset, and a logic circuit known in the art to which the present invention belongs. , Registers, communication modems, data processing devices, and the like.

In addition, the encoding method according to the present specification may be implemented in the form of a computer program written to perform the operations of the encoding apparatus described above and recorded on a computer-readable recording medium.

The computer program is a C / C ++, C #, JAVA, Python that the computer's processor (CPU) can read through the computer's device interface in order for the computer to read the program and execute the methods implemented as the program. It may include a code (Code) coded in a computer language, such as machine language. Such code may include functional code associated with a function or the like that defines the necessary functions for executing the methods, and includes control procedures related to execution procedures necessary for the computer's processor to execute the functions according to a predetermined procedure. can do. In addition, the code may further include memory reference code for additional information or media required for the computer's processor to execute the functions at which location (address address) of the computer's internal or external memory should be referenced. have. Also, if the processor of the computer needs to communicate with any other computer or server remotely in order to execute the functions, the code may be used to communicate with any other computer or server remotely using the communication module of the computer. It may further include a communication related code for whether to communicate, what information or media should be transmitted and received during communication.

The stored medium is not a medium for storing data for a short time such as a register, a cache, a memory, but semi-permanently, and means a medium that can be read by the device. Specifically, examples of the storage medium include, but are not limited to, a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like. That is, the program may be stored in various recording media on various servers to which the computer can access or various recording media on the computer of the user. The media may also be distributed over network coupled computer systems so that the computer readable code is stored in a distributed fashion.

While the embodiments of the present disclosure have been described with reference to the accompanying drawings, a person skilled in the art to which the present disclosure belongs may practice the present disclosure in other specific forms without changing the technical spirit or essential features. I can understand that. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive.

100: Raptor Q Encoder
110: padding
120: intermediate symbol generation unit
121: memory unit
122: calculator
130: LT encoder
200: Raptor Q Decoder
210: padding
220: intermediate symbol generation unit
230: LT decoder

Claims (15)

A padding unit generating an extended source block by adding a padding symbol to a source block including a plurality of source symbols having a predetermined size;
An intermediate symbol generator configured to generate an intermediate symbol using the extended source block; And
An encoding apparatus comprising: an LT encoder for generating a recovery symbol using the intermediate symbol;
The intermediate symbol generation unit,
A memory unit for storing an operation list according to the Gaussian elimination method; And
And an operation unit configured to perform an operation according to the operation list using source symbols generated up to an operation start time for generating the intermediate symbol. The method according to claim 1,
And the operation list includes operation order information, first symbol information to be calculated, second symbol information to be calculated, and arithmetic content information. The method according to claim 2,
And the operation unit stores the operation target list managing the operation order information for which the operation is completed or the operation order information for which the operation is not completed, in the memory unit. The method according to claim 2,
The operation unit may be configured to store, in the memory unit, an unusable symbol list that manages first symbol information and second symbol information that cannot be used in an operation among the first symbol information and the second symbol information included in the operation list. An encoding device. The method according to claim 4,
And the operation unit performs a remaining operation except for an operation including first symbol information or second symbol information included in the unavailable symbol list. The method according to claim 5,
And the operation unit initializes the unavailable symbol list after performing all of the operations operable in the operation list using the source symbols generated up to an operation start point. The method according to claim 1,
And the operation unit performs an operation according to the operation list after a preset number of source symbols is generated. A padding step of generating an extended source block by adding a padding symbol to a source block including a plurality of source symbols having a predetermined size;
Generating an intermediate symbol using the extended source block; And
LT encoding step of generating a repair symbol using the intermediate symbol;
The intermediate symbol generation step,
And performing an operation according to an operation list according to a Gaussian elimination method stored in a memory unit by using the source symbols generated up to an arbitrary operation start time point for generating the intermediate symbol. The method according to claim 8,
And the operation list includes operation order information, first symbol information to be calculated, second symbol information to be calculated, and arithmetic content information. The method according to claim 9,
The operation step further comprises the step of storing in the memory unit arithmetic target list for managing arithmetic operation order information is completed or arithmetic operation information is not completed in the operation list. The method according to claim 9,
The operation may include storing, in the memory unit, an unusable symbol list that manages first symbol information and second symbol information that cannot be used in an operation among the first symbol information and the second symbol information included in the operation list. Encoding method. The method according to claim 11,
The operation may include performing an operation other than an operation including first symbol information or second symbol information included in the unavailable symbol list. The method according to claim 12,
And wherein the calculating step initializes the unavailable symbol list after performing all of the operations that are operable in the operation list using the source symbols generated up to an operation start time. The method according to claim 8,
And wherein the calculating step performs an operation according to the operation list after a preset number of source symbols is generated. A computer program recorded on a computer-readable recording medium, which is written to perform the steps of the encoding method according to any one of claims 8 to 14 in a computer.

Images