Construction method and system of connection matrix of dual-space coupled LDPC

Abstract

The application discloses a connection matrix construction method and system of double space coupling LDPC, comprising the following steps: S1, obtaining an original joint base matrix; S2, performing connection base matrix change on each divided area of the original joint base matrix to construct each new joint base matrix; S3, under an AWGN channel, using different source probabilities and window sizes to evaluate each new joint base matrix by using a JPEXIT algorithm of an SC-LDPC in a JSCC system; and S4, obtaining minimum signal-to-noise ratios SNRs required by each new joint base matrix in the JPEXIT algorithm. By using the technical scheme, the decoding threshold of the double space coupling LDPC code is significantly reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of communication systems, and particularly relates to a connection matrix construction method and system of double space coupled LDPC. BACKGROUND

[0002] In a traditional communication system, source coding and channel coding usually adopt separate design (SSCC), which can simplify modular implementation, but its performance is limited by the Shannon separation theorem, especially in the low complexity or short code length scene, it is difficult to approach the channel capacity. In order to break through this limitation, joint source channel coding (JSCC) has gradually become a research hotspot, which can significantly improve the system robustness through the collaborative design of source and channel. However, the double protograph low-density parity-check code (DP-LDPC) widely used in existing JSCC schemes still has the following problems: the traditional BP decoding needs global iteration, which is difficult to meet the real-time requirement. Therefore, space coupled LDPC codes are introduced into JSCC systems, and SC-LDPC codes use sliding window decoding, which does not need to perform global decoding on the entire received codeword, thereby reducing the delay to meet the real-time requirement. However, using sliding window decoding also sacrifices part of the decoding performance. SUMMARY

[0003] The technical problem to be solved by the application is to provide a connection matrix construction method and system of double space coupled LDPC, which significantly reduces the decoding threshold of double space coupled LDPC codes.

[0004] To achieve the above object, the application adopts the following technical scheme:

[0005] The application provides a connection matrix construction method of double space coupled LDPC, which comprises the following steps:

[0006] Step S1, obtaining an original joint base matrix;

[0007] Step S2, changing the connection base matrix of each divided area of the original joint base matrix to construct each new joint base matrix;

[0008] Step S3, evaluating each new joint base matrix under an AWGN channel;

[0009] Step S4, obtaining the minimum signal-to-noise ratio SNR required by each new joint base matrix in the JPEXIT algorithm.

[0010] Preferably, in step S1, the connection base matrix is selected as a unit connection base matrix, and the original joint base matrix is constructed according to the unit connection base matrix.

[0011] As preferred, in step S2, the original joint base matrix is divided into each divided area as a reference, the connection base matrix of each divided area is changed to obtain a new connection base matrix; the new connection base matrix is combined with the original joint base matrix to replace the original unit connection base matrix to construct each new joint base matrix.

[0012] As preferred, in step S3, the JPEXIT algorithm of SC-LDPC under JSCC system is used to evaluate each new joint base matrix by using different source probabilities and window sizes.

[0013] The application also provides a system for constructing a double space coupled LDPC connection matrix, comprising:

[0014] A first processing module is configured to obtain an original joint base matrix.

[0015] A second processing module is configured to change the connection base matrix of each divided area of the original joint base matrix to construct each new joint base matrix.

[0016] A third processing module is configured to evaluate each new joint base matrix under an AWGN channel.

[0017] A fourth processing module is configured to derive the minimum signal-to-noise ratio (SNR) required by each new joint base matrix in the JPEXIT algorithm.

[0018] As preferred, the first processing module is configured to construct the original joint base matrix according to the unit connection base matrix.

[0019] As preferred, the second processing module is configured to change the connection base matrix of each divided area of the original joint base matrix to obtain a new connection base matrix; the new connection base matrix is combined with the original joint base matrix to replace the original unit connection base matrix to construct each new joint base matrix.

[0020] As preferred, the third processing module is configured to evaluate each new joint base matrix by using the JPEXIT algorithm of SC-LDPC under JSCC system by using different source probabilities and window sizes.

[0021] The double space coupled LDPC code design of the application further optimizes the connection matrix structure by using a dynamic sliding window structure to improve the decoding performance. The original module graph LDPC (P-LDPC) code can use the original module graph external information transfer graph (P-EXIT) algorithm to obtain a decoding threshold and use the decoding threshold as a reference for the performance of the code type design. For the SC-LDPC code under JSCC, the simplified joint P-EXIT (JPEXIT) can be used to perform the performance reference of the code type design of SC-LDPC under JSCC system. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description only constitute a part of the embodiments of the present application, and all other drawings obtained by those skilled in the art without creative effort based on the provided drawings also belong to the protection scope of the present application.

[0023] Figure 1 is a flow chart of a connection matrix construction method of a double space coupled LDPC in the embodiments of the present application;

[0024] Figure 2 is a schematic diagram of a joint base matrix of a double space coupled LDPC under a JSCC system adopted in the embodiments of the present application, taking a window size W=10 as an example;

[0025] Figure 3 is a specific decomposition diagram of a sliding window structure with a window size W=6;

[0026] Figure 4 is an example of a new joint base matrix constructed when the window size W=10;

[0027] Figure 5 is a decoding threshold of different window sizes under different constructed matrices when the source probability Pv=0.01;

[0028] Figure 6 is a decoding threshold of different window sizes under different constructed matrices when the source probability Pv=0.015;

[0029] Figure 7 is a decoding threshold of different window sizes under different constructed matrices when the source probability Pv=0.02. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort belong to the protection scope of the present application.

[0031] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0032] Embodiment 1:

[0033] As Figure 1As shown, the embodiment of the present application provides a connection matrix construction method of double space coupled LDPC, in order to obtain the performance optimization of double space coupled LDPC code, reduce the decoding threshold required in decoding, comprising:

[0034] Step S1, obtaining the original joint base matrix;

[0035] Step S2, changing the connection base matrix of each division area of the original joint base matrix, and constructing each new joint base matrix;

[0036] Step S3, under the AWGN channel, using different source probabilities and window sizes, using the JPEXIT algorithm of SC-LDPC in the JSCC system to evaluate each new joint base matrix;

[0037] Step S4, obtaining the minimum signal-to-noise ratio SNR (decoding threshold) required in the JPEXIT algorithm of each new joint base matrix.

[0038] As an embodiment of the present application, in step S1, the regular code with a source code of 【3,3,3,3】 is used as the source code of source coding, and the regular code with a channel code of 【3,3】 is used as the source code of channel coding, and is taken as an example. As shown in Figure 2 As shown, the source code and the channel code are expanded with m0=2 as the coupling width, the connection base matrix BL1 is initially used as the unit connection base matrix, and the original joint base matrix is constructed. The window size W=10 obtains the size of each base matrix module: the source base matrix BS (10*40), the channel base matrix BC (10*20), the connection base matrix (Bsccv=【0,1】) (10*20), and the joint base matrix size (20*60).

[0039] As an embodiment of the present application, step S2 comprises:

[0040] Step 2.1: as shown in Figure 3 In the joint sliding window process with W=6 as the window size, different characteristics are obtained in the connection base matrix area of each window, different colors are used to represent different windows, and different matrix areas are divided by different shaded parts. For example Figure 3 As shown in the diagonal shadow (overlap) area and the vertical line shadow (unique) area, the two different areas are further divided into (row, column) overlap and (row, column) unique, respectively represented by B_row_overlap (Bro), B_col_overlap (Bco), B_row_single (Brs), and B_col_single (Bcs). Then, the connection matrix is improved for these different areas. Figure 3 Bsccv represents the original connection relationship, including but not limited to the unit connection base matrix.

[0041] Step 2.2: Take the original joint basis matrix in Step 1 as an example, and add an element "1" to each region defined in Step 2.1, to construct a brand new connection basis matrix, which can be obtained as Figure 4 Example of new connection matrix (Bco+1). At the same time, the addition of elements such as column-only, row-only, and more overlapping parts are no longer considered in the preliminary JPEXIT test screening due to poor performance, etc. Further, four new connection basis matrices such as Bco+1, Bco+2, Bro+1, Bco+2 and Bro+1 can be obtained.

[0042] Step 2.3: Combine each brand new connection matrix obtained in Step 2.2 with the source and channel basis matrices involved in Step 1 to obtain each different joint basis matrix.

[0043] Step 3.1: Perform parameter definition in JPEXIT test, each row of the matrix is defined as a check node (i-th CNs), and each column is defined as a variable node (j-th VNs). Source bits s j , are binary Bernoulli sources, where the probability of element "1" appearing is Pv.

[0044] Step 3.2: Use JPEXIT to evaluate the decoding threshold under the AWGN channel. For dual-space coupled LDPC codes, the JPEXIT used is different from that of LDPC under dual-radix graph, which only needs to judge the first window in the sliding window decoding process.

[0045] Step 3.3: Define five different mutual information values (MI).

[0046] I Ev (i,j): Mutual information MI from the j-th VN to the i-th CN.

[0047] I Ec (i,j): Mutual information MI from the i-th CN to the j-th VN.

[0048] I Av (i,j): Mutual information MI from the j-th VN to the i-th CN.

[0049] I Ac (i,j): Mutual information MI from the i-th VN to the j-th CN.

[0050] I APP (j): MI between the posteriori LLR evaluated by the j-th VN and the corresponding source bit s j .

[0051] Mapping the size of the joint basis matrix from step 1 to letter representations, the source basis matrix is ​​(m S *n S The channel basis matrix is ​​(m) C *n C The linking basis matrix is ​​(m) S *n C The joint basis matrix is ​​((m) S +m C )*(n S +n C )).

[0052] Step 3.4: MI update from VNs to CNs:

[0053] For j = 1, ..., m S and i = 1, ..., m S +m C :

[0054]

[0055] Where the functions J(·) and J -1 The simple polynomial approximation of (·) is based on the Gaussian approximation, and the function J BSC Defined as Φ(b i,j ) represents the element b at the corresponding position in the joint basis matrix. i,j The value of .

[0056] I(V;χ) is the MI between VN and χ of the information source. in

[0057] Step 3.5: MI update from VNs to CNs:

[0058] For j = n S +1,…,n S +n c and i = 1, ..., m S +m C :

[0059]

[0060] It is worth noting that Define R = 1 / 2 and E b / N o These are the LDPC code rate and signal-to-noise ratio, respectively.

[0061] For j = 1, ..., n S +n c and i = 1, ..., mS +m C :

[0062] I Ac (i,j)=I Ev (i,j)

[0063] Step 3.6: MI update from CNs to VNs:

[0064] For j=1,…,n S +n C and i=1,…,m S +m C :

[0065]

[0066] and set I Av (i,j)=I Ec (i,j).

[0067] Step 3.7: APP-LLR MI evaluation:

[0068] For j=1,…,n S and i=1,…,m S :

[0069] I APP (j)=J BSC (μ(j),p v )

[0070] where μ(j)=∑ i b i,j [J -1 (I Av (i,j))] 2 . The iteration of MI stops until I APP (j)=1.

[0071] Step 4: Obtain the corresponding signal-to-noise ratio SNR value at this time; wherein, for the double space coupled LDPC code, JPEXIT only needs one window, and for the example of the source code mentioned in step 1, the mutual information needs to be converged only by the first 4 bits. Therefore, the decoding threshold of each new joint base matrix can be obtained.

[0072] The effect of the application can be further illustrated by the following simulation:

[0073] The joint base matrix constructed by each construction example of the application and the unit connection matrix is tested by JPEXIT, and the obtained decoding threshold is compared, as shown in Figure 5 、 Figure 6 、 Figure 7The simulation results prove the effectiveness of the connection matrix construction method of the double space coupled LDPC based on the sliding window overall structure of the application.

[0074] Embodiment 2:

[0075] The embodiment of the application further provides a connection matrix construction system of the double space coupled LDPC, comprising:

[0076] The first processing module is used for acquiring the original joint base matrix.

[0077] The second processing module is used for performing connection base matrix change on each divided area of the original joint base matrix to construct each new joint base matrix.

[0078] The third processing module is used for evaluating each new joint base matrix under the AWGN channel.

[0079] The fourth processing module is used for obtaining the minimum signal-to-noise ratio (SNR) required by each new joint base matrix in the JPEXIT algorithm.

[0080] As an implementation manner of the embodiment of the application, the first processing module is used for constructing the original joint base matrix according to the unit connection base matrix.

[0081] As an implementation manner of the embodiment of the application, the second processing module is used for taking each divided area of the original joint base matrix as a reference to perform connection base matrix change on each divided area to obtain a new connection base matrix; and combining the new connection base matrix with the original joint base matrix to replace the original unit connection base matrix and construct each new joint base matrix.

[0082] As an implementation manner of the embodiment of the application, the third processing module is used for evaluating each new joint base matrix by using the JPEXIT algorithm of the SC-LDPC under the JSCC system with different source probabilities and window sizes.

[0083] The above-described embodiments are only used for describing the preferred modes of the application, and do not limit the scope of the application. Without departing from the design spirit of the application, various modifications and improvements of the technical solutions of the application made by those skilled in the art should fall into the protection scope of the claims of the application.

Claims

1. A method for constructing a connection matrix of a dual-space coupled LDPC, characterized in that, The method comprises the following steps: Step S1, obtaining an original joint basis matrix; Step S2, changing the connection basis matrix of each divided area of the original joint basis matrix to construct each new joint basis matrix; Step S3, evaluating each new joint basis matrix under an AWGN channel; Step S4, obtaining the minimum signal-to-noise ratio SNR required by each new joint basis matrix in the JPEXIT algorithm; In step S1, the unit connection basis matrix is selected, and the original joint basis matrix is constructed according to the unit connection basis matrix; In step S2, the new connection basis matrix is obtained by changing the connection basis matrix of each divided area of the original joint basis matrix; the new connection basis matrix is combined with the original joint basis matrix to replace the original unit connection basis matrix, and each new joint basis matrix is constructed; specifically comprising: Step 2.1: different matrix areas are divided in the connection basis matrix area, different shadow parts, the different areas are further divided into overlapping and unique, and B_row_overlap (Bro), B_col_overlap (Bco), B_row_single (Brs), and B_col_single (Bcs) are used to represent them, respectively; Step 2.2: based on the matrix area division in step 2.1, elements "1" are added to each area defined to construct a new connection basis matrix, and four new connection basis matrices are obtained: Bco plus 1, Bco plus 2, Bro plus 1, and Bco plus 2 and Bro plus 1; Step 2.3: each new connection matrix is combined with the source and channel basis matrix to obtain each different joint basis matrix; In step S3, the JPEXIT algorithm of SC-LDPC in the JSCC system is used to evaluate each new joint basis matrix with different source probabilities and window sizes; wherein, the JPEXIT is used to evaluate the decoding threshold under the AWGN channel, and the first window in the sliding window decoding process is judged.

2. A connection matrix construction system of a dual-space-coupled LDPC, which implements the connection matrix construction method of the dual-space-coupled LDPC of claim 1, characterized by The method comprises the following steps: A first processing module is configured to obtain an original joint basis matrix; A second processing module is configured to change the connection basis matrix of each divided area of the original joint basis matrix to construct each new joint basis matrix; A third processing module is configured to evaluate each new joint basis matrix under an AWGN channel; A fourth processing module is configured to obtain the minimum signal-to-noise ratio SNR required by each new joint basis matrix in the JPEXIT algorithm.

Images