KR101206434B1 - method for decoding using dynamic scheduling scheme for low density parity check codes and apparatus thereof - Google Patents

Abstract

The present invention relates to a decoding method and apparatus therefor using a dynamic scheduling technique for a low density parity check code.

In the present invention, the low-density parity check code is sequentially decoded. In particular, the messages are scheduled in the order of a large difference between before and after updating of the message transmitted from the variable node to the check node. As a result, it shows superior performance compared to the existing dynamic scheduling scheme, and unnecessary decoding and reordering processes are eliminated.

Low density parity test, LDPC, decoding,

Description

Method for decoding using dynamic scheduling scheme for low density parity check codes and apparatus

The present invention relates to a decoding method, and more particularly, to a decoding method and apparatus therefor using a dynamic scheduling technique for a low density parity check code.

The present invention is derived from the research conducted as part of the IT source technology development project of the Ministry of Knowledge Economy and the Ministry of Information and Communication Research and Development. [Task management number: 2008-F-002-01, Task name: Next-generation tactical defense communication source technology development] .

In order to satisfy the high data rate of the next generation mobile communication system, researches are being conducted on a coding scheme having an effective performance with a short length and a fast convergence rate decoding scheme, in particular, a low density parity check codes. (Hereinafter referred to as LDPC code) has been actively studied for various decoding methods.

The decoding method of LDPC code is basically a check node or a variable node to receive an error from a predetermined number of neighbor nodes, update the message, and return the updated message back to the neighbor node. Decrypt while correcting. The output message of a variable node or check node is a function of all messages coming into that node except the line on which the message is to be sent, i.e. the message coming from the edge.

Determining the update order of such messages is called scheduling. Scheduling decoding includes a static scheduling decoding method for updating messages sequentially according to a fixed order, and a dynamic scheduling decoding method for dynamically updating messages.

Static scheduling decoding schemes are classified into SBP (Shuffled Belief Propagation) and LBP (Layered Belief Propagation) according to whether they are updated in the variable node direction or in the check node direction. Since the introduction of SBP and LBP, various modified forms of techniques have been studied.

The existing dynamic scheduling decoding method, also called RBP (residual belief propagation), is a method of first updating a message having the largest change amount before and after updating among messages transmitted from a check node to a variable node.

The difference in the message value before and after the update is close to 0 indicates that the message is almost converged. On the contrary, the difference in the message value before and after the update indicates that the message is not converged. Therefore, the convergence speed of the dynamic scheduling decoding method that updates the message that has not yet converged first is much faster than the decoding convergence speed of the static scheduling decoding methods such as the existing SBP and LBP which are updated in a fixed order. In addition, the dynamic scheduling decoding method can effectively solve the trapping set error, and thus has an improved performance compared to the static scheduling decoding method.

However, the conventional dynamic scheduling decoding scheme for the LDPC code performs scheduling in the order of the difference in the values before and after the update of the message delivered from the check node to the variable node. The delivered message has a disadvantage of low reliability because it is based on only one check expression. In particular, since there is a high possibility that a message may be misjudged, there is a problem in that an incorrectly determined message is updated first and a new error occurs.

An object of the present invention is to provide a decoding method and apparatus using a more reliable dynamic scheduling technique for LDPC code.

Another object of the present invention is to provide a decoding method and apparatus using a dynamic scheduling technique, which is low in complexity and easy to implement.

In the decoding method according to an aspect of the present invention, in the method of defining a plurality of check nodes and variable nodes from a parity check matrix and performing decoding based on the check nodes, the sorting reference values of all variable-check messages transmitted from the variable node to the check node are determined. Selecting a variable-check message having the largest alignment reference value by comparison, wherein the selected variable-check message is passed from the first variable node to the first check node; Updating a check-variable message passed from the first check node to a second variable node for all second variable nodes connected to the first check node; And calculating, for all second check nodes connected to the second variable node, an alignment reference value of the variable-check message transmitted from the second variable node to the second check node based on the updated check-variable message. It includes.

Finding a first variable node coupled to the selected variable-check message; And updating, at the first variable node, all variable-check messages sent to all first check nodes connected to the first variable node.

The decoding apparatus according to another aspect of the present invention, in the apparatus for defining a plurality of check nodes and variable nodes from the parity check matrix and performing decoding based on the parity check matrix, initializes the variable-check messages transmitted from the variable node to the check node, An initialization unit for initializing check-variable messages transmitted from the check node to the variable node; A check node updating unit for updating variable-check messages transmitted from the variable node to the check node; A variable node updater for updating check-variable messages transmitted from the check node to the variable node; And a processing unit for comparing the sorting reference values of all the variable-check messages and selecting the variable-checking message having the largest sorting reference value and transmitted from the first variable node to the first check node. Here, the variable node updater updates the check-variable message transmitted from the first check node to the second variable node for all second variable nodes connected to the first check node under the control of the processor. And the check node updating unit arranges the variable-check message transmitted from the second variable node to the second check node based on the updated check-variable message for all second check nodes connected to the second variable node. Calculate the reference value.

According to the present invention, the decoding method is faster than the existing dynamic scheduling scheme and the reliability thereof is improved, thereby providing excellent decoding performance. Therefore, stable data transmission is possible.

In addition, complexity can be significantly reduced by avoiding unnecessary computation and reordering. In addition, as the complexity of the decoding process is reduced, high-speed communication is also guaranteed. In addition, it can be applied not only to the communication field but also to various next generation systems requiring error correction technology.

In addition, by selecting a variable node that includes edges with a large difference in message values before and after updating, and updating all edges connected to the selected variable node at once, the number of selections of the decoding process is reduced, thereby more effectively reducing the complexity of the decoding process. It is possible to maintain excellent decoding performance.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Throughout the specification, when a part is said to "include" a certain component, it means that it can further include other components, without excluding other components unless specifically stated otherwise. In addition, the terms “… unit”, “… group” described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software.

Hereinafter, embodiments of the present invention will be described in more detail with reference to the accompanying drawings.

The decoding method according to an embodiment of the present invention is performed using a parity check matrix and a Tanner graph, and the parity check matrix may be represented by a Tanner graph. As many check nodes are created as the number of rows, and as many variable nodes as the number of columns are created to form the Tanner graph. If the (i, j) element of the matrix is 1, the i th check node and the j th variable node are connected by edges to become neighboring nodes.

Iterative decoding of the LDPC code is achieved through repeated message transfer between the variable node and the check node. If the check node c _i and the variable node v _j are neighboring nodes, the messages can be generated as follows.

[Equation 1]

&Quot; (2) "

는 변수 노드 v_j 에서 체크 노드 c_i 로 전송되는 메시지의 업데이트 함수이며, 이를 변수-체크 메시지라고 명명할 수 있다.

는 체크 노드 c_i 에서 변수 노드 v_j 로 전송되는 메시지의 업데이트 함수이며, 이를 체크-변수 메시지라고 명명할 수 있다.here,

Is the update function of the message sent from the variable node v _j to the check node c _i , which can be called a variable-check message.

Is the update function of the message sent from the check node c _i to the variable node v _j , which can be called a check-variable message.

는 변수 노드 v_j 에 대한 채널 정보이며,

와 같이 나타낼 수 있다. 여기서,

는 수신 신호를 나타낸다. Also,

Is the channel information for the variable node v _j ,

As shown in Fig. here,

Indicates a received signal.

는 j번째 변수 노드에 연결된 모든 체크 노드들 중에서 i번째 체크 노드를 제외한 체크 노드들을 나타낸다. Also,

Denotes check nodes except the i th check node among all the check nodes connected to the j th variable node.

는 체크 노드(c_a)들에 연결된 모든 변수 노드들 중에서 j번 째 변수 노드를 제외한 변수 노드들을 나타낸다.

Denotes the variable nodes except the j th variable node among all the variable nodes connected to the check nodes c _a .

Before describing a decoding method according to an embodiment of the present invention using an LDPC having such a feature, a general decoding method will be described.

In the general decoding method of the LDPC code, the check-variable message is updated using the value of the variable-check message obtained in the (i-1) -th iteration during the i-th iteration. Then all variable-check messages are updated using the check-variable message newly obtained in the previous step. This process is repeated until the stop condition is satisfied. Here, the variable-check message may be represented by Equation 1 above, and the check-variable message may be represented by Equation 2.

In particular, the conventional dynamic scheduling technique is a method of first updating a message having the largest sort criterion value called residual, which is called RBP (residual belief propagation). The alignment reference value is calculated from the difference between the values before and after the update of the check-variable message, and can be expressed as follows.

 &Quot; (3) "

는 정렬 기준값이며,

는 업데이트에 의하여 새로이 산출된 체크-변수 메시지의 값이다. here,

Is the sort criterion,

Is the value of the check-variable message newly calculated by the update.

1 is an exemplary diagram illustrating a decoding process according to a dynamic scheduling scheme of a conventional LDPC code.

As shown in FIG. 1, the decoding process according to the general dynamic scheduling technique, that is, the decoding process of the RBP, may be represented in three steps. In FIG. 1, variable nodes and check nodes are represented by circles and rectangles, respectively.

가 선택되었다고 하자. 선택된 체크-변수 메시지

의 값이 크게 변화하므로, 해당 변수 노드가 다른 값으로 결정될 확률이 높다. 그러므로 해당 변수 노드에 대한 체크-변수 메시지를 먼저 업데이트하면 다른 노드를 바르게 결정하는데 도움이 되어 전체 복호 성능을 향상시킬 수 있다. 따라서 가장 큰 값의 정렬 기준값을 가지는 체크-변수 메시지를 먼저 선택하는 것이다. In the first step ((a) of FIG. 1), the difference between the values before and after the update is obtained for the check-variable messages, respectively, and the pair of the check node and the variable node having the largest value, that is, the largest alignment criterion value, , Check-variable message sent over the edge connecting these two nodes

Let is chosen. Selected Check-Variable Message

Because the value of changes significantly, there is a high probability that the variable node will be determined to a different value. Therefore, updating the check-variable message for that variable node first can help determine the other node correctly and improve the overall decoding performance. Therefore, the check-variable message with the largest sort criterion is selected first.

를 업데이트 하고 대응되는 에지를 처리 대상에서 제외시킨다. 즉 체크 노드(c_i)와 변수 노드(v_j)에 연결된 에지를 통하여 전송되는 체크-변수 메시지의 정렬 기준값을 0으로 처리하여

, 현재의 반복 복호 과정에서 더 이상 선택되지 않도록 한다. Then, the selected message

Is updated and the corresponding edge is excluded from processing. That is, the sorting reference value of the check-variable message transmitted through the edge connected to the check node (c _i ) and the variable node (v _j ) is treated as 0.

In the current iterative decoding process, it is no longer selected.

를 업데이트 한다. 즉, j번째 변수 노드에 연결되어 있는 체크 노드들 중에서 상기에서 제외 처리 된 에지에 연결된 i번째 체크 노드(c_i)를 제외한 모든 체크 노드들(c_a)에 대하여, 변수-체크 메시지

의 값을 산출한다. In the next second step (FIG. 1B), all

Update That is, for all the check nodes c _a except for the i th check node c _i connected to the excluded edge among the check nodes connected to the j th variable node, the variable-check message

Calculate the value of.

에 대한 정렬 기준값(

)들을 업데이트한다. 체크-변수 메시지들에 대한 업데이트 전후의 값들의 차이를 토대로 스케줄링이 이루어지는데 체크 노드들(c_a)에서 변수 노드(v_b)들로 전달되는 메시지가 업데이트 되었기 때문에, 이러한 메시지들에 대응하는 에지들에 대한 정렬 기준값들을 새로이 산출해 주는 단계가 수행되어야 한다. And in the third step (Fig. 1 (c)), all of the sorting reference values to be used in the next decoding process,

Sort by for

). Scheduling is based on the difference in the values before and after the update to the check-variable messages, since the message passed from the check nodes c _a to the variable nodes v _b has been updated, the edge corresponding to these messages. A step of newly calculating the sorting reference values for these fields should be performed.

)을 산출한다. Specifically, based on the values for the variable-check messages obtained in the second step above, among the variable nodes connected to the check nodes c _a , the j th variable node v _j connected to the excluded edge is The sorting criterion value (the difference value before and after updating the check-variable messages for the excluded variable nodes v _b )

) Is calculated.

And reorder the check-variable messages except the messages that have not yet been selected, ie the messages that were selected in the first step above, according to their sorting criteria. For example, check-variable messages may be sorted in ascending order of sort criteria.

Through these steps, all messages (or edges) are updated only once in one iterative decoding process.

Although the existing RBP decoding method having such a feature is an effective decoding method having a fast decoding convergence speed, it is necessary to improve error correction performance and complexity.

Specifically, the existing RBP decoding method calculates the alignment reference value from the change of the message value based on only one check expression. That is, as shown in Equation 2, the alignment reference value is calculated based on the check expression for the variable nodes connected to one check node. As a result, in the RBP decoding method, which is a greedy algorithm, when an incorrectly determined message has the largest alignment reference value, a new error that does not occur in the static scheduling decoding method may be generated and propagated.

)이 산출될 때 이미 메시지 값

이 산출되었음에도 불구하고 다시

이 중복 산출되는 불필요한 단계가 발생한다. In addition, in the conventional decoding method, when the message to be sent from the variable node to the check node is updated, one decoding process is completed, whereas the RBP decoding method calculates the sorting reference value based on the check-variable message. An additional step is needed to update the message passed to the variable node. In particular, when the check-variable message is updated, the sort criteria (

Is already a message value when

Despite this was calculated again

This redundant calculation takes place.

There is also the hassle of unnecessarily rearranging the unselected messages Q each time the sorting reference values of each message are calculated. Where Q is sorted by the size of the sort criterion and represents a set of messages that have not yet been updated.

Accordingly, an embodiment of the present invention is to provide a decoding method using a dynamic scheduling technique of LPDC code that does not perform an unnecessary process while reducing an error occurrence rate. In particular, the embodiment of the present invention uses the difference between the values before and after the update of the message transmitted from the variable node to the check node as an alignment reference value for scheduling.

2 is an exemplary diagram illustrating a decoding process according to the first embodiment of the present invention, and FIG. 3 is a flowchart of the decoding method according to the first embodiment of the present invention for performing the decoding process according to FIG. 2. The decoding method according to the first embodiment of the present invention may also be referred to as a variable-to-check RBP (ie, VCRBP method).

= 0) 하고, 또한 모든 변수-체크 메시지들의 값을 제2 값(여기서는

이며, 이는 채널값을 나타낸다.)으로 초기화(

) 한다(S100~S110). First, as shown in FIG. 3, the value of all check-variable messages is initialized to a first value (here “0”).

= 0), and also the value of all variable-check messages to a second value (here

, Which indicates the channel value.)

(S100 ~ S110).

를 0으로 초기화 시켜준다 (S120). 단, 이때에는 해당 메시지를 업데이트 대상에서 제외시키지 않는다. 초기화 다음에는 최초 업데이트 될 변수-체크 메시지(

) 를 임의로 정해준다. 예를 들어, 연결된 체크 노드의 수가 가장 많은 변수 노드의 변수-체크 메시지 중 하나를 선택한다. 그러나 그 이후부터는 가장 큰

에 대응되는

를 선택한다 (S130).In the initial decoding process, there is no previous decoding process, so all

Initializes to 0 (S120). In this case, however, the message is not excluded from the update target. After initialization, the variable-check message to be updated first (

) Is arbitrarily set. For example, select one of the variable-check messages of the variable node with the largest number of connected check nodes. But since then the biggest

Corresponding to

Select (S130).

) 처리 대상에서 제외시킨다 (S140). 도 2의 (a)에서 처리 대상에서 제외된 메시지를 점선으로 표시하였다. Set the sort criterion value of the selected variable-check message to 0 (

) To be excluded from the process (S140). In (a) of FIG. 2, a message excluded from the processing target is indicated by a dotted line.

를 갱신한다. 즉, j번째 체크 노드에 연결되어 있는 변수 노드들 중에서 i번째 변수 노드는 제외한 모든 변수 노드들(v_a)에 대하여, 체크-변수 메시지들의 값

들을 산출한다 (S150). And all

Update the. That is, for all the variable nodes v _a except the i th variable node among the variable nodes connected to the j th check node, the value of the check-variable messages

To calculate them (S150).

를 갱신한다(S160). 즉, 변수 노드들(v_a)에 연결되어 있는 체크 노드들 중에서 제외 처리된 j번째 체크 노드를 제외한 체크 노드들(c_a)에 대한 변수-체크 메시지들의 값

들을 산출한다. Next, as in the second step (b) of FIG.

Update (S160). That is, the value of the variable-check messages for the check nodes c _a except the j th check node excluded from the check nodes connected to the variable nodes v _a .

Calculate them.

)을 산출하고, 모든 변수-체크 메시지들의 정렬 기준값들 중에서 가장 큰 정렬 기준값을 찾는다 (S180).After that, it is determined whether the setting condition for ending the decoding process is satisfied (S170). If the setting condition is not satisfied, the alignment is a difference value before and after updating based on the values of the variable-check messages obtained in step S160. Reference value (

), And find the largest sort criterion value among the sort criterion values of all the variable-check messages (S180).

The sorting reference value of the variable-check messages according to an embodiment of the present invention may be expressed as follows.

&Quot; (4) "

는 업데이트에 의하여 새로이 산출된 변수-체크 메시지의 값을 나타낸다. here,

Denotes the value of the variable-check message newly calculated by the update.

As in Equation 2 above, the check-variable message is calculated based on the check expression related to only one check node, but the variable-check message is related to all the check nodes except for one, as in Equation 1 Calculated based on Therefore, in the embodiment of the present invention, the entire code can be more accurately decoded based on the alignment reference value obtained from the message having higher reliability, that is, the variable-check message.

들을 구한 다음에, 구해진

들을 서로 비교한다. 결과에 따라 위의 단계 (S130~S170)를 반복 수행한다.Specifically, all of the variable-check messages are based on Equation 4 described above.

After saving them,

Compare them to each other. Repeat the above steps (S130 ~ S170) according to the result.

If the setting condition is satisfied, the decoding process is terminated (S190).

According to the first embodiment of the present invention as described above, the alignment reference value is calculated from the variable-check messages based on the check expressions related to all the check nodes except one, so that the entire code is more reliably obtained. It can be decoded.

회의 메시지(또는 에지)를 선택하는 과정이 필요하게 된다. Meanwhile, the decoding method according to the first embodiment of the present invention as described above selects the message (or edge) having the largest difference before and after updating the message. Because of this, one message at a time, i.e., an edge, is selected, and all edges must be selected in order to complete one decoding process. For example, if the total number of variable nodes is N and the degree representing the average number of edges of each variable node is d _v , to complete the entire decoding process,

The process of selecting a conference message (or edge) is required.

Accordingly, the second embodiment of the present invention describes a decoding method that can reduce the process of selecting a message while performing decoding as in the first embodiment.

In the second embodiment of the present invention, when the variable-check message having the largest alignment criterion value is selected, delivery of the message occurs all at once through all edges connected to the corresponding variable node. As a result, the number of variable-check message selections made in one iterative decoding process is reduced to N times.

Specifically, the scheduling is basically based on selecting a pair of variable nodes and check nodes having a large difference in message values before and after the update, and a large difference in values before and after the update of the variable-check message is determined in the previous decoding process. The variable node has been updated, which means that the message value from that variable node has changed significantly. Therefore, if the variable-check message sent through the edge connected to the corresponding variable node has been selected with the largest alignment criterion, the variable-check messages to be delivered through the other edge connected to the variable node are also likely to have a large alignment reference. Is likely to be selected. Therefore, in the second embodiment of the present invention, when the edge having the largest alignment reference value, that is, the variable-check message is selected, the variable-check messages transmitted through all other edges connected to the variable node corresponding to the message are also updated. .

4 is an exemplary diagram illustrating a decoding process according to the second embodiment of the present invention, and FIG. 5 is a flowchart of a decoding method according to the second embodiment of the present invention for performing the decoding process according to FIG. 4. The decoding method according to the second embodiment of the present invention may also be referred to as a node-wise VCRBP, that is, an NVCRBP method.

= 0) 하고, 또한 모든 변수-체크 메시지들의 값을 제2 값(여기서는 "

")으로 초기화(

) 한다 (S300~S310). First, as shown in FIG. 5, the value of all check-variable messages is initialized to a first value (here “0”).

= 0), and also sets the value of all variable-check messages to a second value (here "

To ""

(S300 ~ S310).

를 0으로 초기화 시켜준다 (S320). 단, 이때에는 해당 메시지를 업데이트 대상에서 제외시키지 않는다. 초기화 다음에는 최초 업데이트 될 변수-체크 메시지(

) 를 임의로 정해준다. 예를 들어, 연결된 체크 노드의 수가 가장 많은 변수 노드의 변수-체크 메시지 중 하나를 선택한다. 그러나 그 이후부터는 가장 큰

에 대응되는

를 선택한다 (S330).In the initial decoding process, there is no previous decoding process, so all

Initializes to 0 (S320). In this case, however, the message is not excluded from the update target. After initialization, the variable-check message to be updated first (

) Is arbitrarily set. For example, select one of the variable-check messages of the variable node with the largest number of connected check nodes. But since then the biggest

Corresponding to

Select (S330).

를 갱신한다. 즉, 위의 단계 (S330)에서 찾은 변수 노드(v_i)에 연결되어 있는 모든 체크 노드들(c_a)들로 전송될 변수-체크 메시지들(

)을 업데이트한다. 그리고 변수-체크 메시지들(

)들의

값을 0으로 하여 처리 대상에서 제외시킨다 (S340). 이와 같이 본 발명의 제2 실시 예에서는 선택된 변수 노드(v_i)에 에지로 연결되어 있는 모든 체크 노드들(c_a)들에 대한 변수-체크 메시지들을 모두 업데이트하고 처리 대상에서 제외시킴으로써, 다음 반복 단계에서 변수 노드(v_i)에 연결된 또 다른 에지가 선택되지 않도록 한다. 그 결과, 한 번의 반복 과정 내에 이루어지는 에지 선택의 회수를 감소시켜 복호 속도를 보다 향상시킬 수 있다. And all

Update the. That is, the variable-check messages to be transmitted to all the check nodes c _a connected to the variable node v _i found in step S330 above (

). And variable-check messages (

)field

A value of 0 is excluded from the processing target (S340). As described above, according to the second embodiment of the present invention, all the variable-check messages for all the check nodes c _a connected to the edge of the selected variable node v _i are updated and excluded from the processing. Ensure that no other edge connected to the variable node v _i is selected in the step. As a result, the decoding speed can be further improved by reducing the number of edge selections made in one iteration process.

를 갱신한다. 즉, 업데이트된

값을 기반으로 체크 노드(c_a)들에 에지로 연결되어 있는 변수 노드들 중에서 선택된 i번째 변수 노드(v_i)를 제외한 변수 노드(v_b)들로 보내지는 체크-변수 메시지들

의 값을 산출한다 (S350).As shown in FIG. 4 (a) below, all

Update the. That is, updated

Check-variable messages sent to the variable nodes (v _b ) except the i th variable node (v _i ) among the variable nodes edged to the check nodes (c _a ) based on the value.

Calculate the value of (S350).

즉, 모든 변수 노드(v_b)들에 연결되어 있는 체크 노드들 중에서 선택된 변수 노드(v_i)에 에지로 연결되어 있는 체크 노드들(c_a)을 제외한 모든 체크 노드들(c_b)에 대한, 변수-체크 메시지들

를 산출한다 (S360). Thereafter, as shown in FIG. 4B and FIG. 5,

That is, for all the check nodes c _b except for the check nodes c _a connected to the edge of the selected variable node v _i among the check nodes connected to all the variable nodes v _b . , Variable-checked messages

To calculate (S360).

를 토대로 업데이트 전후의 값의 차이인 정렬 기준값

을 구하고, 모든

들을 서로 비교한다 (S380). 결과에 따라 위의 단계(S330~S370)를 반복 수행한다.After that, it is determined whether the setting condition for ending the decoding process is satisfied (S370), and when the setting condition is not satisfied, it is obtained in step S360.

Based on the difference in the values before and after the update

Save all,

Compare them to each other (S380). Repeat the above steps (S330 ~ S370) according to the result.

If the setting condition is satisfied, the decoding process is terminated (S390).

Decoding methods according to embodiments of the present invention as described above may be implemented through a decoding device having the following structure.

A decoding apparatus according to an embodiment of the present invention performs decoding on a received signal based on decoding methods according to an embodiment of the present invention using a parity check matrix and a Tanner graph. The received received signal may be processed as a codeword of length n to be decoded. The test matrix may be represented as a Tanner graph. As many check nodes are created as the number of rows, and as many variable nodes as the number of columns are created to form the Tanner graph. If the (i, j) element of the matrix is 1, the i th check node and the j th variable node are connected by edges to become neighboring nodes.

6 is a structural diagram of a decoding apparatus according to an embodiment of the present invention.

Decoding apparatus according to an embodiment of the present invention, as shown in Figure 6, the initialization unit for initializing the messages delivered to all the check nodes from all variable nodes, and initializes the messages delivered to all the variable nodes from all check nodes ( 10), the check node update unit 20 for updating the messages transmitted from the variable node to the check node, the variable node update unit 30 for updating the messages transferred from the check node to the variable node, for scheduling The reference value calculator 40 calculates a difference between the sorting reference values, that is, the values before and after the update of the variable-check messages, selects a predetermined variable-check message based on the calculated sorting reference values, and accordingly, checks node updating unit 20. And a processing unit 50 for operating the variable node updating unit 30 to perform decoding. 20, a variable node is an update result of the update unit 30 may further include a memory 60 to be stored.

The processor 50 selects the variable-check message having the largest difference between the values before and after the update based on the decoding methods as described above, and operates the check node update unit 20 and the variable node update unit 30 based on the result. The decoding is performed by updating other check nodes connected to the edge corresponding to the selected variable-check message or updating all variable-check messages connected to the variable node corresponding to the selected variable-check message.

As a result of simulation by applying the decoding method according to the embodiments of the present invention to the decoding device as described above, the following results were obtained.

7 is a graph showing the performance of the decoding method (VCRBP) according to the first embodiment of the present invention compared with the conventional decoding methods (BP, LBP, RBP). In particular, FIG. 7 is a graph showing a frame error rate (FER) according to each method when the code length 576 and the code rate 1/2 of the IEEE 802.16e standard are used and the maximum number of repetitions is fixed to eight. to be. Referring to FIG. 7, it can be seen that the decoding method (VCRBP) according to the first embodiment of the present invention shows a performance improvement of about 0.3 dB at FER 10 ^-4 compared to the best RBP among the existing decoding methods. have.

8 is a graph illustrating the performance of the decoding methods (VCRBP, NVCRBP) according to the first and second embodiments of the present invention compared with the conventional decoding methods (BP, LBP, RBP). In particular, Figure 8 is a graph showing the FER according to each method when using a code length 576, code rate 1/2 of the IEEE 802.16e standard and fixed the maximum number of repetitions to eight. Referring to FIG. 8, the decoding method NVCRBP according to the second embodiment of the present invention has a low complexity even though the number of comparison of the reference values is reduced, and the decoding method according to the first embodiment of the present invention The performance is similar to that of VCRBP), and compared to the existing methods, the performance is excellent.

9 is a diagram illustrating an approximate continuous firing amount of the decoding method VCRBP according to the first embodiment of the present invention in comparison with the conventional decoding methods BP, LBP, and RBP. In FIG. 9, since the complexity of the check-variable message function corresponding to Equation 2 is much greater than that of the variable-check message function corresponding to Equation 1 above, in comparing the calculation amounts of the respective decoding methods, In the method, the approximate amount of calculation is calculated by the number of times the check-variable message is calculated. Referring to FIG. 9, it can be seen that the calculation amount of the decoding method VCRBP according to the first embodiment of the present invention is smaller than that of the conventional decoding method RBP.

FIG. 10 illustrates the performance of the decoding method VCRBP according to the first embodiment of the present invention in comparison with the existing decoding methods BP, LBP, and RBP when the same amount of calculation is performed based on the calculation amount formula of FIG. 9. It is a graph. In FIG. 10, FER performance was compared by using a code length 576 and a code rate 1/2 of the IEEE 802.16e standard and fixing the maximum number of repetitions to 228, 12, and 38, respectively. Referring to FIG. 10, it can be seen that the decoding method VCRBP according to the first embodiment of the present invention has excellent performance compared to the same amount of calculation.

The embodiments of the present invention described above are not only implemented by the apparatus and method but may be implemented through a program for realizing the function corresponding to the configuration of the embodiment of the present invention or a recording medium on which the program is recorded, The embodiments can be easily implemented by those skilled in the art from the description of the embodiments described above.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It belongs to the scope of right.

1 is a diagram illustrating a decoding process using an existing dynamic scheduling scheme of an LDPC code.

2 is a diagram illustrating a decoding process according to a first embodiment of the present invention.

3 is a flowchart illustrating a decoding method according to a first embodiment of the present invention.

4 is a diagram illustrating a decoding process according to a second embodiment of the present invention.

5 is a flowchart illustrating a decoding method according to a second embodiment of the present invention.

6 is a diagram showing the structure of a decoding apparatus according to an embodiment of the present invention.

7 is a graph showing the performance of the decoding method according to the first embodiment of the present invention in comparison with the conventional decoding methods.

8 is a graph showing the performance of the decoding method according to the first and second embodiments of the present invention in comparison with the conventional decoding methods.

9 is a diagram showing an approximate continuous firing amount of the decoding method according to the first embodiment of the present invention in comparison with conventional decoding methods.

FIG. 10 is a graph illustrating performance of a decoding method according to the first embodiment of the present invention compared with conventional decoding methods when having the same calculation amount based on the calculation amount formula of FIG. 9.

Claims (15)

In a method of defining a plurality of check nodes and variable nodes from a parity check matrix and performing decoding based on the check nodes, Compares the sorting reference values of all variable-check messages passed from the variable node to the check node to select the variable-check message having the largest sorting reference value, wherein the selected variable-check message is passed from the first variable node to the first check node. Delivered; Updating a check-variable message passed from the first check node to a second variable node for all second variable nodes connected to the first check node; And Calculating, for all second check nodes connected to the second variable node, an alignment criterion value of the variable-check message transmitted from the second variable node to the second check node based on the updated check-variable message. Decoding method comprising a. The method of claim 1, The selected variable-check message is a message transferred from one first variable node to one first check node, Updating the check-variable message updates the check-variable messages delivered to all the second variable nodes in the one first check node. 3. The method of claim 2, And initializing the sorting reference value of the selected variable-check message to a set value and excluding it from processing. The method of claim 1 Finding a first variable node coupled to the selected variable-check message; And Updating, at the first variable node, all variable-check messages sent to all first check nodes connected to the first variable node Decoding method further comprising. 5. The method of claim 4, And initializing the sorting reference values of the variable-check messages sent to all the first check nodes connected to the first variable node to a set value, and excluding them from the processing object. delete The method according to claim 3 or 5 The decoding method repeats the steps sequentially a plurality of times, In subsequent iterations And in the forwarding step, a variable-check message having the largest sorting reference value among remaining variable-check messages is selected except for the variable-check message in which the sorting reference value is initialized. delete 6. The method according to any one of claims 1 to 5, Prior to the step of delivering, Initiating variable-check messages passed from the variable node to the check node; And Initializing check-variable messages passed from the check node to the variable node Decoding method further comprising. The method of claim 9 And initialize the variable-check messages to a value corresponding to a received channel characteristic of the received signal and initialize the check-variable messages to zero. In an apparatus for defining a plurality of check nodes and variable nodes from a parity check matrix and performing decoding based on the check nodes, An initialization unit for initializing variable-check messages transmitted from the variable node to the check node and initializing check-variable messages transferred from the check node to the variable node; A check node updating unit for updating variable-check messages transmitted from the variable node to the check node; A variable node updater for updating check-variable messages transmitted from the check node to the variable node; And A processing unit for comparing the variable reference values of all the variable-check messages and selecting the variable-check message having the largest alignment reference value and transmitted from the first variable node to the first check node Including, The variable node updater updates the check-variable message transmitted from the first check node to the second variable node for all second variable nodes connected to the first check node, under the control of the processor. The check node updating unit arranges the variable-check message transmitted from the second variable node to the second check node based on the updated check-variable message for all second check nodes connected to the second variable node. To calculate the reference value, Decoding device. The method of claim 11, The processing unit finds a first variable node connected to the selected variable-check message, And the check node updating unit updates all of the variable-check messages transmitted from the first variable node to all the first check nodes connected to the first variable node. The method of claim 12, And the variable node updater updates check-variable messages respectively transmitted from all first check nodes to all second variable nodes. The method of claim 12, The processing unit, Initialize the sorting reference value of the selected variable-check message to a setting value, or initialize the sorting reference values of the variable-checking messages transmitted to all the first check nodes connected to the first variable node to the setting value and exclude them from processing. , Decoding device. The method according to any one of claims 11 to 14. And the sorting reference value is an absolute value of a value obtained by subtracting a value after the variable-check message is updated from a value before the variable-check message is updated.

Images