LDPC (Low Density Parity Check) code soft decision decoding device
By adopting a unified control method to manage data input and output in the LDPC code soft decision decoder, data routing is simplified, and the problem of high hardware resource consumption caused by the complex data routing network in the prior art is solved, and efficient parallel translation is achieved code and cost-reducing effect.
Patent Information
- Application Number
- CN202410950719.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-16
- Publication Date
- 2024-11-19
Smart Images

Figure CN118984158A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of communication technology, and in particular to an LDPC code soft decision decoding device. Background Art
[0002] LDPC (Low-density Parity-check) code is a low-density parity-check code. It is an error-correcting code type widely used in communication systems and has excellent error correction capabilities. In 4G and 5G mobile communication networks and WiFi local area networks, LDPC codes are selected as the error correction code scheme for data channels. In the fields of satellite high-speed data transmission and deep space communication, LDPC codes are also the main error correction code style used in satellite-to-ground links. Therefore, designing a low-complexity LDPC decoder is of great significance for reducing the power consumption and cost of communication equipment and expanding the application areas of LDPC codes.
[0003] At present, LDPC codes are decoded in an iterative manner, and the decoding algorithms are mainly divided into two categories, soft decision algorithms and hard decision algorithms. The hard decision decoding algorithm demodulates the received data to obtain a binary bit sequence, and then performs a decoding operation on the obtained binary sequence. The hard decision decoding algorithm is simple to calculate and has low storage overhead, but the decoding operation has low utilization of channel information and its error correction capability is relatively poor. The advantage of the soft decision decoding algorithm over the hard decision decoding algorithm is that the decoder input soft information represented by the log-likelihood ratio describes the probability of each bit of demodulated data taking 0 or 1, that is, the data confidence, which reflects the impact of the channel on each transmitted bit. Compared with the hard decision decoding algorithm, it greatly improves the utilization of channel information, thereby improving the decoding error correction capability.
[0004] In order to reduce the computational complexity of the soft decision decoder and improve the convergence speed of iterative decoding, a variety of algorithms such as the belief propagation algorithm, the minimum sum algorithm, and the TDMP layered decoding algorithm have been proposed. The design and implementation of the LDPC code soft decision decoder adopts one of the above algorithms. At the same time, in order to improve the throughput of the LDPC code soft decision decoder and reduce processing delays, parallel operations are usually used inside the decoder. The operation unit processes multiple input soft information each time in a multi-input multi-output manner.
[0005] However, in the actual design and implementation of LDPC code soft decision decoder, the LDPC decoder based on the belief propagation algorithm involves addition and division operations, and the variable node update and check node update are divided into two independent processes, requiring multiple variable node calculation units and check node calculation units. The data interaction between the variable node calculation unit and the check node calculation unit is complex, and a large number of data selectors are needed to build a complex data routing network. As the parallelism of the decoder calculation increases, the complexity of the data routing network will increase sharply, resulting in large hardware resource overhead and high power consumption of the decoder. At the same time, the intermediate data generated during the decoding process also requires a large amount of internal storage space, further increasing the consumption of hardware resources. The LDPC decoder based on the minimum sum algorithm uses arithmetic operations such as addition and subtraction that are easy to implement in hardware to construct variable node update units and check node update units, which reduces the computational complexity of the decoder. However, the data interaction between the variable node calculation unit and the check node calculation unit is complex, and a large number of data selectors are still required to build a complex data routing network. As the parallelism of the decoder calculation increases, the complexity of the data routing network will increase sharply, resulting in large hardware resource overhead and high power consumption of the decoder. At the same time, the intermediate data generated during the decoding process also requires a large amount of internal storage space, further increasing the consumption of hardware resources. The iterative decoding operation of the LDPC decoder based on the TDMP layered decoding algorithm is performed in layers. After completing the update of a layer of check nodes, the corresponding variable nodes are directly updated with the new information, and then the update of the next layer of messages can directly use the updated information of the previous layer, which can reduce the message transmission delay, avoid the long-term caching of a large amount of decoding intermediate information, and improve the iterative convergence speed. However, when the LDPC decoder based on the TDMP layered decoding algorithm performs parallel decoding operations, there is a problem of complex reading and writing of decoding soft information and posterior probability information. It is also necessary to build a complex data routing network between the decoding operation unit and the data storage unit to ensure that the decoding units working in parallel can accurately obtain the data involved in the decoding. For example, for the LDPC decoding adopted by the IEEE802.11ad protocol, when the parallelism of the decoding operation unit is 16, according to the existing design scheme of the LDPC decoder based on the TDMP layered decoding algorithm, a fully interconnected data write routing network with 16 input ports and 52 output ports, and a fully interconnected data write routing network with 52 input ports and 16 output ports need to be built between the operation unit and the posterior probability information storage unit, and the hardware implementation cost is huge. Summary of the invention
[0006] In order to solve part or all of the technical problems existing in the above-mentioned prior art, the present invention provides an LDPC code soft decision decoding device.
[0007] The technical solution of the present invention is as follows:
[0008] An LDPC code soft decision decoding device is provided, the device comprising:
[0009] A decoding soft information storage unit, wherein the input end of the decoding soft information storage unit serves as the input end of the LDPC code soft decision decoding device and is also connected to the output end of the second adder, and the output end of the decoding soft information storage unit is respectively connected to the input end of the first adder and the input end of the decoding decision unit, the decoding soft information storage unit is used to receive and store the soft information input to the LDPC code soft decision decoding device, to receive and store the soft information after iterative decoding input by the second adder, to output the updated soft information to the decoding decision unit after completing a round of iterative decoding, and to output the received soft information obtained by the previous round of iterative decoding to the first adder when performing iterative decoding, wherein when performing the first iterative decoding, the decoding soft information storage unit outputs the received soft information input to the LDPC code soft decision decoding device as the soft information obtained by the previous round of iterative decoding to the first adder;
[0010] The first adder, the input end of the first adder is also connected to the output end of the posterior probability information storage unit, the output end of the first adder is respectively connected to the input end of the hierarchical decoding DEC unit and the input end of the second adder, the first adder is used to add the received soft information to the received posterior probability information after taking the negative, and output the addition result to the hierarchical decoding DEC unit and the second adder respectively;
[0011] The hierarchical decoding DEC unit, the output end of the hierarchical decoding DEC unit is respectively connected to the input end of the posterior probability information storage unit and the input end of the second adder, the hierarchical decoding DEC unit is used to perform a decoding operation on the received information to obtain updated posterior probability information, and output the posterior probability information to the posterior probability information storage unit and the second adder;
[0012] The a posteriori probability information storage unit is used to receive and store the a posteriori probability information, and is used to output the a posteriori probability information to the first adder when performing iterative decoding, wherein when performing the first iterative decoding, all elements in the a posteriori probability information output by the a posteriori probability information storage unit to the first adder are initialized to 0;
[0013] The second adder is used for adding the received addition result to the received posterior probability information after negating it to obtain updated soft information, and outputting the soft information to the decoding soft information storage unit;
[0014] The decoding decision unit is used to determine whether the set iteration stop criterion is met according to the received soft information, and if so, output the decoding result; if not, enable the LDPC code soft decision decoding device to continue iterative decoding based on the current iterative decoding result.
[0015] In some optional implementations, the decoding soft information storage unit includes: a first data input selection circuit, a first dual-port memory, and a first data output selection circuit;
[0016] The first data input selection circuit is provided with a multiplexer, a plurality of first input branches, a plurality of second input branches and a plurality of output branches, one end of the plurality of first input branches is respectively connected to the output end of the multiplexer, the input end of the multiplexer is used as the input end of the decoding soft information storage unit, and is used to receive the soft information input into the LDPC code soft decision decoding device, one end of the plurality of second input branches is used to be connected to the second adder, and the other end of the first input branch and the other end of the second input branch are switchably connected to one end of the output branch through a single-pole double-throw switch;
[0017] The first dual-port memory includes a plurality of first dual-port memories, the plurality of first dual-port memories are respectively connected to the other ends of the plurality of output branches of the first data input selection circuit, and each of the first dual-port memories is configured with a read address generator and a write address generator;
[0018] The first data output selection circuit is provided with multiple input ports and multiple output ports, the inner ends of the multiple input ports are respectively connected to the inner ends of the multiple output ports through control switches, the outer ends of the multiple input ports are respectively connected to the multiple first dual-port memories, and the outer ends of the multiple output ports are used to connect to the first adder and the decoding decision unit.
[0019] In some optional implementations, the layered decoding DEC unit receives information Q m,n , for receiving information Q m,n The decoding operation includes the following:
[0020] Calculate information Q m,n The absolute value of
[0021] Determine the absolute value |Q m,n |The minimum value min1 and the second minimum value min2 of each column, and record the row index corresponding to the minimum value min1 of each column;
[0022] The absolute value |Q m,n |The elements of each column are filled with the minimum value min1 of the column and the element with the row number Index of the column is replaced with the second minimum value min2 of the column, and the matrix V is obtained. m,n ;
[0023] Judge and record information Q m,n The sign of each element in is obtained according to the sign of each element. m,n );
[0024] The symbol matrix sign(Q m,n ) multiply each column element by one, and add the multiplication result to the symbol matrix sign(Q m,n ) Multiply the elements of each column separately to get the new symbol matrix C m,n ;
[0025] According to R m,n =S·V m,n ·C m,n Calculate the posterior probability information R m,n And output, where S represents the preset normalization coefficient.
[0026] In some optional implementations, the hierarchical decoding DEC unit includes: an absolute value calculation module, a symbol determination module, a data symbol operation module, a data sorting module, a data filling module, a data replacement module and a data symbol recovery module;
[0027] The absolute value calculation modules include multiple modules, the symbol determination modules include multiple modules, the input ends of the multiple absolute value calculation modules are respectively connected to the input ends of the multiple symbol determination modules and serve as the input end of the hierarchical decoding DEC unit, the output ends of the multiple absolute value calculation modules are respectively connected to the input end of the data sorting module, the output ends of the multiple symbol determination modules are respectively connected to the input end of the data symbol operation module, the input end of the data filling module is connected to the input end of the data sorting module, the output end of the data sorting module is connected to the input end of the data replacement module, the output end of the data replacement module and the output end of the data symbol operation module are respectively connected to the input end of the data symbol recovery module, and the output end of the data symbol recovery module serves as the output end of the hierarchical decoding DEC unit;
[0028] The plurality of absolute value calculation modules and the plurality of sign determination modules are used to receive in parallel in the same clock cycle a plurality of data of the same column of information input into the hierarchical decoding DEC unit, and respectively calculate and output the absolute values and positive and negative signs of the plurality of data;
[0029] The data symbol operation module is used to receive the positive and negative symbols input by the plurality of symbol determination modules, multiply the positive and negative symbols, and multiply the multiplication result by the positive and negative symbols respectively to obtain and output a column of elements of the symbol matrix;
[0030] The data filling module is used to generate multiple data serial numbers and sequentially fill them into multiple absolute values output by multiple absolute value calculation modules;
[0031] The data sorting module is used to sort the received multiple absolute values, determine the minimum value among the multiple absolute values and its corresponding data sequence number and the second smallest value among the multiple absolute values and its corresponding data sequence number, and output the received multiple absolute values, the determined minimum value and its corresponding data sequence number and the second smallest value and its corresponding data sequence number;
[0032] The data replacement module is used to replace the minimum value of multiple absolute values with the next smallest value and output it, and replace other values of the multiple absolute values with the minimum value and output it according to the data sequence number corresponding to the minimum value and the data sequence number corresponding to the next smallest value;
[0033] The data symbol recovery module is used to multiply each column element received from the data symbol operation module, each column element received from the data replacement module and a preset normalization coefficient to obtain and output posterior probability information.
[0034] In some optional implementations, the posterior probability information storage unit includes: a second data input selection circuit, a second dual-port memory, and a second data output selection circuit;
[0035] The second data input selection circuit is provided with a plurality of input ports and a plurality of output ports, the inner ends of the plurality of input ports are respectively connected to the inner ends of the plurality of output ports through control switches, and the outer ends of the plurality of input ports are used to be connected to the hierarchical decoding DEC unit;
[0036] The second dual-port memory includes a plurality of second dual-port memories, the plurality of second dual-port memories are respectively connected to the external ends of the plurality of output ports of the second data input selection circuit, each of the second dual-port memories is configured with a read address generator and a write address generator, and the read address generator and the write address generator configured in each of the second dual-port memories are connected to an address offset controller, and the address offset controller is capable of generating parameters for address generation;
[0037] The second data output selection circuit is provided with a plurality of input ports and a plurality of output ports, the inner ends of the plurality of input ports of the second data output selection circuit are respectively connected to the inner ends of the plurality of output ports of the second data output selection circuit through control switches, the outer ends of the plurality of input ports of the second data output selection circuit are respectively connected to a plurality of the second dual-port memories, and the outer ends of the plurality of output ports of the second data output selection circuit are used to be connected to the first adder.
[0038] In some optional implementations, the decoding decision unit includes: a plurality of register groups, a decoding verification module and a verification decision module;
[0039] The multiple register groups are respectively connected to the decoding soft information storage unit and the decoding verification module, each register in the multiple register groups is used to store single-bit decoding hard decision data, the decoding verification module has multiple output ports, each output port of the decoding verification module corresponds to a row of the LDPC code check matrix, the multiple output ports of the decoding verification module are respectively connected to the input end of the check decision module, the output end of the check decision module is connected to the multiple register groups, the check decision module is used to judge the results output by the multiple output ports of the decoding verification module, and select the first multiple bits cached in the register group as the decoding result output when the judgment is passed, and enable the LDPC code soft decision decoding device to continue the next round of iterative decoding based on the current iterative decoding result when the judgment is not passed.
[0040] In some optional implementations, the device further includes: a control unit, the control unit being configured to store decoding control parameters and perform corresponding control according to the stored decoding control parameters;
[0041] The decoding control parameters include: control signals of the decoding soft information storage unit and the posterior probability information storage unit for the data input selection circuit and the data output selection circuit in the hierarchical decoding stage, initialization addresses of the read address generator and the write address generator of the decoding soft information storage unit and the posterior probability information storage unit in the hierarchical decoding stage, and address offset control amount of the posterior probability information storage unit in the hierarchical decoding stage.
[0042] The main advantages of the technical solution of the present invention are as follows:
[0043] The LDPC code soft decision decoding device of the present invention can achieve the maximum computational parallelism, significantly improve the decoder throughput, and reduce the decoding delay; and the data input and output of the decoding soft information storage unit and the posterior probability information storage unit adopt a unified control method, do not include a complex fully interconnected data routing network, and only control the data through switch switching, which can greatly reduce the data access complexity in the parallel decoding process, simplify the decoder structure, and reduce costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] The drawings described herein are used to provide a further understanding of the embodiments of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0045] Figure 1A schematic diagram of the structure of an LDPC code soft decision decoding device according to an embodiment of the present invention;
[0046] Figure 2 A schematic diagram of the working process of an LDPC code soft decision decoding device according to an embodiment of the present invention;
[0047] Figure 3 A schematic diagram of the structure of a decoding soft information storage unit according to an embodiment of the present invention;
[0048] Figure 4 A flowchart of a decoding operation of a layered decoding DEC unit according to an embodiment of the present invention;
[0049] Figure 5 A schematic diagram of the structure of a hierarchical decoding DEC unit according to an embodiment of the present invention;
[0050] Figure 6 A schematic diagram of the structure of a posterior probability information storage unit according to an embodiment of the present invention;
[0051] Figure 7 The figure is a schematic diagram of the structure of a decoding decision unit according to an embodiment of the present invention. DETAILED DESCRIPTION
[0052] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0053] The technical solution provided by the embodiments of the present invention is described in detail below with reference to the accompanying drawings.
[0054] See also Figure 1 An embodiment of the present invention provides an LDPC code soft decision decoding device, the LDPC code soft decision decoding device comprising:
[0055] A decoding soft information storage unit, wherein the input end of the decoding soft information storage unit is used as the input end of the LDPC code soft decision decoding device and is also connected to the output end of the second adder, and the output end of the decoding soft information storage unit is respectively connected to the input end of the first adder and the input end of the decoding decision unit, and the decoding soft information storage unit is used to receive and store soft information input to the LDPC code soft decision decoding device, to receive and store soft information after iterative decoding input by the second adder, to output the updated soft information to the decoding decision unit after completing a round of iterative decoding, and to output the received soft information obtained by the previous round of iterative decoding to the first adder when performing iterative decoding, wherein when performing the first iterative decoding, the decoding soft information storage unit outputs the received soft information of the input LDPC code soft decision decoding device as the soft information obtained by the previous round of iterative decoding to the first adder;
[0056] a first adder, wherein the input end of the first adder is also connected to the output end of the posterior probability information storage unit, the output end of the first adder is respectively connected to the input end of the hierarchical decoding DEC unit and the input end of the second adder, and the first adder is used to add the received soft information after negating it to the received posterior probability information, and output the addition result to the hierarchical decoding DEC unit and the second adder respectively;
[0057] A hierarchical decoding DEC unit, wherein the output end of the hierarchical decoding DEC unit is respectively connected to the input end of the posterior probability information storage unit and the input end of the second adder, and the hierarchical decoding DEC unit is used to perform a decoding operation on the received information to obtain updated posterior probability information, and output the posterior probability information to the posterior probability information storage unit and the second adder;
[0058] a posterior probability information storage unit, used to receive and store the posterior probability information, and to output the posterior probability information to the first adder when performing iterative decoding, wherein when performing the first iterative decoding, all elements in the posterior probability information output by the posterior probability information storage unit to the first adder are initialized to 0;
[0059] A second adder, configured to add the negative of the received addition result to the received posterior probability information to obtain updated soft information, and output the soft information to a decoding soft information storage unit;
[0060] The decoding decision unit is used to determine whether the set iteration stop criterion is met according to the received soft information. If so, the decoding result is output; if not, the LDPC code soft decision decoding device continues iterative decoding based on the current iterative decoding result.
[0061] refer to Figure 1-2, the LDPC code soft decision decoding device provided by one embodiment of the present invention adopts the TDMP decoding algorithm to decompose each round of iterative decoding into multiple layers of decoding operations to complete. Specifically, when the LDPC code soft decision decoding device has no input, the LDPC code soft decision decoding device remains in an idle state. When the LDPC code soft decision decoding device receives soft information, that is, when the soft information input enable is valid, the LDPC code soft decision decoding device enters the soft information input state, and the input soft information is received and stored by the decoding soft information storage unit; when the soft information input is completed, the LDPC code soft decision decoding device automatically enters the layered decoding state and starts the layered decoding operation. Taking the nth layer decoding operation in the mth round of iterative decoding as an example, the decoding soft information storage unit will receive the soft information L used to participate in the nth layer decoding operation in the soft information updated by the received m-1th round of iterative decoding. m-1,n The output is sent to the first adder, and the a posteriori probability information storage unit receives the a posteriori probability information updated by the m-1th round of iterative decoding and uses it to participate in the nth layer decoding operation. m-1,n Output to the first adder, the first adder adds the soft information L m-1,n After taking the negative and the posterior probability information R m-1,n Add and get the result Q m,n =R m-1,n +(-L m-1,n ), and the addition result Q m,n The outputs are respectively sent to the hierarchical decoding DEC unit and the second adder. The hierarchical decoding DEC unit receives the information Q m,n Perform decoding operation to obtain the posterior probability information R updated by iterative decoding m,n =DEC(Q m,n ), DEC(·) represents the decoding operation, and the updated posterior probability information R m,n The posterior probability information storage unit receives and stores the posterior probability information R m,n The second adder adds the information Q input by the first adder m,n After taking the negative, the posterior probability information R of the layered decoding DEC unit input m,n Add to get the soft information L updated by iterative decoding m,n =R m,n +(-Q m,n ), and update the soft information L m,n Output to the decoding soft information storage unit, the decoding soft information storage unit receives and stores the soft information L m,n. After completing a round of iterative decoding through layered decoding operation, the decoding soft information storage unit outputs the stored updated soft information to the decoding decision unit. The decoding decision unit determines whether the set iteration stop criterion is met based on the received soft information. The iteration stop criterion can be that the decoding decision unit passes the verification or the number of iterative decoding reaches the maximum number of iterations. If so, the decoding decision unit outputs the decoding result and completes the decoding. If not, the LDPC code soft decision decoding device continues the next round of iterative decoding based on the current iterative decoding result until the set iterative stop criterion is met, outputs the decoding result, and completes the decoding. When the decoding result is output, the LDPC code soft decision decoding device returns to the idle state and waits for new soft information input.
[0062] Furthermore, assuming that the LDPC code length is K, the code rate is r, the submatrix dimension of its check matrix is J×J, the submatrix is a zero matrix or a permutation matrix of the identity matrix cyclically right shifted by several columns, based on the TDMP decoding algorithm, the number of decoding operation layers is K(1-r) / J. Based on the above assumptions, the posterior probability information R m,n and soft information m,n They are all two-dimensional matrices, and the number of matrix columns is J, and the number of matrix rows is the number of non-zero submatrices K contained in the LDPC code check matrix at the nth layer. n .
[0063] In one embodiment of the present invention, when executing each layer of decoding operation, according to the characteristics of the LDPC code check matrix, part of the soft information and the posterior probability information are selected for decoding operation, and the calculated data is used to update the soft information and the posterior probability information, so that when executing a new layer of decoding operation, the updated data of the previous layers of decoding can be directly used, thereby accelerating the decoding convergence speed.
[0064] refer to Figure 3 , To achieve the above functions, in one embodiment of the present invention, the decoding soft information storage unit specifically includes: a first data input selection circuit, a first dual-port memory and a first data output selection circuit;
[0065] The first data input selection circuit is provided with a multiplexer, a plurality of first input branches, a plurality of second input branches and a plurality of output branches, one end of the plurality of first input branches is respectively connected to the output end of the multiplexer, the input end of the multiplexer is used as the input end of the decoding soft information storage unit, and is used to receive the soft information of the input LDPC code soft decision decoding device, one end of the plurality of second input branches is used to be connected to the second adder, and the other end of the first input branch and the other end of the second input branch are switchably connected to one end of the output branch through a single-pole double-throw switch;
[0066] The first dual-port memory comprises a plurality of first dual-port memories, the plurality of first dual-port memories are respectively connected to the other ends of the plurality of output branches of the first data input selection circuit, and each first dual-port memory is configured with a read address generator and a write address generator;
[0067] The first data output selection circuit is provided with a plurality of input ports and a plurality of output ports, the inner ends of the plurality of input ports are respectively connected to the inner ends of the plurality of output ports through control switches, the outer ends of the plurality of input ports are respectively connected to the plurality of first dual-port memories, and the outer ends of the plurality of output ports are used to be connected to the first adder and the decoding decision unit.
[0068] Further, assuming that the LDPC code length is K, the code rate is r, the dimension of the submatrix constituting its check matrix is J×J, the submatrix is a zero matrix or a permutation matrix of the unit matrix cyclically right-shifted by several columns, based on the TDMP decoding algorithm, the number of decoding operation layers is K(1-r) / J. In view of the above assumptions, in one embodiment of the present invention, the first dual-port memory is set to K / J, the storage depth of the first dual-port memory is set to J, the number of first input branches, the number of second input branches and the number of output branches of the first data input selection circuit are set to K / J, and the number of input ports and the number of output ports of the first data output selection circuit are set to K / J.
[0069] In one embodiment of the present invention, based on the decoding soft information storage unit of the above-mentioned structural type, each first dual-port memory is configured with a read address generator and a write address generator, which can ensure that different memories can read and write data independently, and the first data input selection circuit supports the time-sharing writing of the input soft information and the soft information updated by the layered decoding into the first dual-port memory for caching. When the input soft information needs to be cached, under the control of the multiplexer, the input soft information is sequentially written into the first first dual-port memory to the K / Jth first dual-port memory. When executing the nth layer of decoding, K first dual-port memories are selected from the K / J first dual-port memories according to the position of the non-zero submatrix in each layer of the check matrix. n The first dual-port memory reads data, K n The number of non-zero submatrices contained in the LDPC code check matrix at the nth layer is closed, and the corresponding control switch in the first data output selection circuit is closed to output the read data. The soft information updated by the layered decoding is written into the first dual-port memory and the storage address before the update in the original location storage mode.
[0070] refer to Figure 4 In one embodiment of the present invention, the hierarchical decoding DEC unit is based on the information Q input by the first adder m,n Get the posterior probability information R m,n The decoding operation DEC(·) performed specifically includes the following contents:
[0071] Calculate information Q m,n The absolute value of
[0072] Determine the absolute value |Q m,n |The minimum value min1 and the second minimum value min2 of each column, and record the row index corresponding to the minimum value min1 of each column;
[0073] The absolute value |Q m,n |The elements of each column are filled with the minimum value min1 of the column and the element with the row number Index of the column is replaced with the second minimum value min2 of the column, and the matrix V is obtained. m,n ;
[0074] Judge and record information Q m,n The sign of each element in is obtained according to the sign of each element. m,n );
[0075] The symbol matrix sign(Q m,n ) multiply each column element by one, and add the multiplication result to the symbol matrix sign(Q m,n ) Multiply the elements of each column separately to get the new symbol matrix C m,n ;
[0076] According to R m,n =S·V m,n ·C m,n Calculate the posterior probability information R m,n And output, where S represents the preset normalization coefficient.
[0077] Further, based on the above-defined decoding operation process, refer to Figure 5 , attached Figure 5 In the figure, Sign represents a sign determination module, |·| represents an absolute value calculation module. In one embodiment of the present invention, in order to implement the above decoding operation, the hierarchical decoding DEC unit includes: an absolute value calculation module, a sign determination module, a data sign operation module, a data sorting module, a data filling module, a data replacement module and a data sign recovery module;
[0078] The absolute value calculation modules include multiple modules, the symbol determination modules include multiple modules, the input ends of the multiple absolute value calculation modules are respectively connected to the input ends of the multiple symbol determination modules and serve as the input end of the hierarchical decoding DEC unit, the output ends of the multiple absolute value calculation modules are respectively connected to the input end of the data sorting module, the output ends of the multiple symbol determination modules are respectively connected to the input end of the data symbol operation module, the input end of the data filling module is connected to the input end of the data sorting module, the output end of the data sorting module is connected to the input end of the data replacement module, the output end of the data replacement module and the output end of the data symbol operation module are respectively connected to the input end of the data symbol recovery module, and the output end of the data symbol recovery module serves as the output end of the hierarchical decoding DEC unit;
[0079] Multiple absolute value calculation modules and multiple symbol determination modules are used to receive multiple data of the same column of information input into the hierarchical decoding DEC unit in parallel in the same clock cycle, and respectively calculate and output the absolute values and positive and negative signs of the multiple data; the data symbol operation module is used to receive the positive and negative signs input by the multiple symbol determination modules, multiply the positive and negative signs, and multiply the multiplication results with the positive and negative signs respectively, to obtain and output a column of elements of the symbol matrix; the data filling module is used to generate multiple data sequence numbers and fill them into the multiple absolute values output by the multiple absolute value calculation modules in turn; the data sorting module is used to sort the multiple absolute values received and determine the minimum value among the multiple absolute values and its corresponding data sequence number and the next smallest value among the multiple absolute values and its corresponding data sequence number, and output the received multiple absolute values, as well as the determined minimum value and its corresponding data sequence number and the next smallest value and its corresponding data sequence number; the data replacement module is used to replace the minimum value among the multiple absolute values with the next smallest value and output it, and replace the other values of the multiple absolute values with the minimum value and output it according to the data sequence number corresponding to the minimum value and the data sequence number corresponding to the next smallest value; the data symbol recovery module is used to multiply the column elements input by the received data symbol operation module, the column elements input by the received data replacement module and the preset normalization coefficient to obtain and output the posterior probability information.
[0080] Specifically, taking the n-th layer decoding operation in the m-th round of iterative decoding as an example, K in the multiple absolute value calculation modules n K in the absolute value calculation module and multiple sign determination modules n The symbol determination modules receive Q m,n K in the same column n data, the absolute value calculation module determines and outputs the absolute value of the corresponding data, and the sign determination module determines and outputs the positive and negative signs of the corresponding data; n At the same time, the data filling module generates K absolute values for the input data sorting module. n data sequence number (for example, the values are 0, 1, ..., K n-1) Input the data sorting module together, K n The data numbers are filled into K in sequence n absolute values, that is, an absolute value and a data sequence number are combined into one data, and the data filled by the data filling module is the maximum unsigned number under the given data bit width, which does not affect the sorting of the valid input data; the data sorting module sorts the received multiple absolute values, determines the minimum value and its corresponding data sequence number and the second smallest value and its corresponding data sequence number, and outputs the received multiple absolute values, the determined minimum value and its corresponding data sequence number and the second smallest value and its corresponding data sequence number to the data replacement module; the data replacement module is used to replace the minimum value of the multiple absolute values with the second smallest value according to the data sequence number corresponding to the minimum value and the data sequence number corresponding to the second smallest value, and replace the other values of the multiple absolute values with the minimum value, and output the replaced multiple absolute value data to the data symbol recovery module; at the same time, in the process of data processing by the data sorting module and the data replacement module, the data symbol operation module also multiplies each positive and negative sign according to the positive and negative signs input by the received multiple sign determination module, and multiplies the multiplication results with each positive and negative sign respectively to obtain the symbol matrix C m,n The data symbol recovery module multiplies the received column elements input by the data symbol operation module, the received elements input by the data replacement module and the preset normalization coefficient to obtain the posterior probability information R m,n Based on the above process, as time goes by, the data symbol recovery module can obtain the posterior probability information R m,n The elements of each column determine and output the posterior probability information R m,n .
[0081] Among them, due to the existence of the data filling module, the hierarchical decoding DEC unit can adapt to K n Any parallelism within the range of 2 to J without adjusting the circuit structure of the data sorting module.
[0082] Further, refer to Figure 6 ,To achieve the above functions, in one embodiment of the present invention, the posterior probability information storage unit includes: a second data input selection circuit, a second dual-port memory, and a second data output selection circuit;
[0083] The second data input selection circuit is provided with a plurality of input ports and a plurality of output ports, the inner ends of the plurality of input ports are respectively connected to the inner ends of the plurality of output ports through control switches, and the outer ends of the plurality of input ports are used to be connected to the hierarchical decoding DEC unit;
[0084] The second dual-port memory includes a plurality of second dual-port memories, the plurality of second dual-port memories are respectively connected to the external ends of the plurality of output ports of the second data input selection circuit, each second dual-port memory is configured with a read address generator and a write address generator, and the read address generator and the write address generator configured in each second dual-port memory are both connected to an address offset controller, and the address offset controller can generate parameters for address generation;
[0085] The second data output selection circuit is provided with a plurality of input ports and a plurality of output ports. The inner ends of the plurality of input ports of the second data output selection circuit are respectively connected to the inner ends of the plurality of output ports of the second data output selection circuit through control switches, the outer ends of the plurality of input ports of the second data output selection circuit are respectively connected to a plurality of second dual-port memories, and the outer ends of the plurality of output ports of the second data output selection circuit are used to be connected to the first adder.
[0086] Further, assuming that the LDPC code length is K, the code rate is r, the dimension of the submatrix constituting its check matrix is J×J, the submatrix is a zero matrix or a permutation matrix of the unit matrix cyclically right-shifted by several columns, based on the TDMP decoding algorithm, the number of decoding operation layers is K(1-r) / J. In view of the above assumptions, in one embodiment of the present invention, the second dual-port memory is set to K / J, and the storage depth of the i-th second dual-port memory is J·η i , η i The number of times the i-th first dual-port memory of the decoding soft information storage unit participates in the hierarchical decoding operation in each round of iterative decoding, the number of input ports and the number of output ports of the second data input selection circuit, and the number of input ports and the number of output ports of the second data output selection circuit are all set to K / J.
[0087] In one embodiment of the present invention, based on the a posteriori probability information storage unit of the above-mentioned structural type, the a posteriori probability information storage unit and the decoding soft information storage unit have the same control mode for the data output selection circuit, and both select the dual-port memory with the same serial number to participate in the hierarchical decoding operation according to the check matrix. In a certain round of iterative decoding, if the i-th second dual-port memory in the a posteriori probability information storage unit is l-th (l=1,2,…,η i ) participates in the hierarchical decoding operation, then in the same clock cycle, when the read address of the i-th first dual-port memory in the decoding soft information storage unit is w L When , the read address of the i-th second dual-port memory in the posterior probability information storage unit will be configured as w accordingly L +(l-1)J, the function of the address frequency shift controller is to generate parameter l for address generation. The posterior probability information updated by the layered decoding operation is written into the second dual-port memory and the storage address before updating in the original address storage mode.
[0088] Further, refer to Figure 7 ,To achieve the above functions, in one embodiment of the present invention, the decoding and decision unit includes: a plurality of register groups, a decoding and verification module, and a verification and decision module;
[0089] A plurality of register groups are respectively connected to a decoding soft information storage unit and a decoding check module, each register in the plurality of register groups is used to store single-bit decoding hard decision data, the decoding check module has a plurality of output ports, each output port of the decoding check module corresponds to a row of an LDPC code check matrix, the plurality of output ports of the decoding check module are respectively connected to an input end of a check decision module, the output end of the check decision module is connected to a plurality of register groups, the check decision module is used to make a decision on the output results of the plurality of output ports of the decoding check module, and when the decision is passed, the first plurality of bits cached in the register group are selected as the decoding result output, and when the decision is not passed, the LDPC code soft decision decoding device is enabled to continue the next round of iterative decoding based on the current iterative decoding result.
[0090] Furthermore, assuming that the LDPC code length is K, the code rate is r, the dimension of the submatrix constituting its check matrix is J×J, the submatrix is a zero matrix or a permutation matrix of the unit matrix cyclically right-shifted by several columns, based on the TDMP decoding algorithm, the number of decoding operation layers is K(1-r) / J. For the above assumptions, the number of register groups is K / J, the length of the register group is J, the decoding check module contains K(1-r) output ports, and when the judgment is passed, the check decision module selects the first K·r bits cached in the register group as the decoding result output.
[0091] In one embodiment of the present invention, when the last layer of decoding operation is executed in each round of iterative decoding, the decoding decision unit can be used to determine whether the decoding result meets the iteration stop criterion and output the decoding result. Specifically, based on the decoding decision unit of the above-mentioned structural type, when the last layer of decoding operation begins to be executed, the dual-port memory that does not participate in the current decoding operation in the decoding soft information storage unit begins to read the stored soft information in sequence, and sends it to the decoding decision unit after hard judgment; at the same time, the soft information updated by the last layer of decoding operation also sends its hard decision result to the decoding decision unit when it returns to the decoding soft information storage unit; the storage method of these hard decision results in K / J register groups is consistent with the order of the corresponding soft information in the decoding soft information storage unit. When all K hard decision results are written, they are output in parallel and sent to the decoding verification module. The decoding check module includes K (1-r) output ports, each of which corresponds to a row of the LDPC code check matrix. Among the K elements in the same row, the input data corresponding to the element "1" is simultaneously connected to the output port. After the XOR operation, these data are used as the output data of the port; the check judgment module judges the output results of the K (1-r) output ports of the decoding check module. When all are 0, the check passes, and the first K·r bits cached in the register group are selected as the decoding result output. When the check judgment fails, the LDPC code soft decision decoding device continues the next round of iterative decoding based on the current iterative decoding result.
[0092] Furthermore, based on the structural type of each unit in the LDPC code soft decision decoding device set above, in one embodiment of the present invention, the LDPC code soft decision decoding device also includes: a control unit, the control unit is used to store decoding control parameters and perform corresponding control according to the stored decoding control parameters.
[0093] Specifically, the decoding control parameters include: control signals of the decoding soft information storage unit and the posterior probability information storage unit for the data input selection circuit and the data output selection circuit in the hierarchical decoding stage, initialization addresses of the read address generator and the write address generator of the decoding soft information storage unit and the posterior probability information storage unit in the hierarchical decoding stage, and address offset control amount of the posterior probability information storage unit in the hierarchical decoding stage.
[0094] Specifically, assuming that the coding length of the LDPC code is K, the code rate is r, the dimension of the submatrix that constitutes its check matrix is J×J, the submatrix is a zero matrix or a permutation matrix of the unit matrix cyclically right shifted by several columns, based on the TDMP decoding algorithm, the number of decoding operation layers is K(1-r) / J.
[0095] Based on the above assumptions, the pattern of the control signal of the data input selection circuit and the data output selection circuit of the decoding soft information storage unit and the posterior probability information storage unit in the hierarchical decoding stage is set to a two-dimensional array C1 containing K(1-r) / J rows and K / J columns. The elements of the two-dimensional array C1 take values of 0 and 1, where 0 indicates that the switch is disconnected and 1 indicates that the switch is closed. When performing the i-th layer decoding operation, the i-th row of the two-dimensional array C1 is used to control the data input selection circuit and the data output selection circuit of the decoding soft information storage unit and the posterior probability information storage unit.
[0096] Based on the above assumptions, the initialization address pattern of the read address generator and the write address generator of the decoding soft information storage unit and the posterior probability information storage unit in the hierarchical decoding stage is set to a two-dimensional array C2 containing K(1-r) / J rows and K / J columns. The values of the elements of the two-dimensional array C2 are the cyclic shift values corresponding to the submatrix at the corresponding position in the LDPC code check matrix. In particular, if the submatrix corresponding to an element is a zero matrix, the element in C2 is set to 0. When performing the i-th layer decoding operation, the i-th row of the two-dimensional array C2 is used to initialize the read address generator and the write address generator in the decoding soft information storage unit and the posterior probability information storage unit.
[0097] Based on the above assumptions, the address offset control amount of the posterior probability information storage unit in the hierarchical decoding stage is set to a two-dimensional array C3 containing K(1-r) / J rows and K / J columns. The value range of the elements in the i-th row and j-th column of C3 is 0 to K(1-r) / J-1, indicating the number of hierarchical decoding times that the j-th first dual-port memory in the decoding soft information storage unit has participated in before executing the i-th layer hierarchical decoding. When executing the i-th layer decoding operation, the K / J elements in the i-th row of the two-dimensional array C3 are the offsets output by the K / J address offset controllers in the posterior probability information storage unit.
[0098] The LDPC code soft decision decoding device provided in one embodiment of the present invention can achieve the maximum computational parallelism (i.e., the dimension of the check matrix submatrix), significantly improve the decoder throughput, and reduce the decoding delay; moreover, the data input and output of the decoding soft information storage unit and the posterior probability information storage unit adopt a unified control method, do not include a complex fully interconnected data routing network, and only control the data through switch switching, which can greatly reduce the data access complexity in the parallel decoding process, simplify the decoder structure, and reduce costs.
[0099] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A LDPC code soft decision decoding device, characterized in that: The device comprises: A decoding soft information storage unit, wherein the input end of the decoding soft information storage unit serves as the input end of the LDPC code soft decision decoding device and is also connected to the output end of the second adder, and the output end of the decoding soft information storage unit is respectively connected to the input end of the first adder and the input end of the decoding decision unit, the decoding soft information storage unit is used to receive and store the soft information input to the LDPC code soft decision decoding device, to receive and store the soft information after iterative decoding input by the second adder, to output the updated soft information to the decoding decision unit after completing a round of iterative decoding, and to output the received soft information obtained by the previous round of iterative decoding to the first adder when performing iterative decoding, wherein when performing the first iterative decoding, the decoding soft information storage unit outputs the received soft information input to the LDPC code soft decision decoding device as the soft information obtained by the previous round of iterative decoding to the first adder; The first adder, the input end of the first adder is also connected to the output end of the posterior probability information storage unit, the output end of the first adder is respectively connected to the input end of the hierarchical decoding DEC unit and the input end of the second adder, the first adder is used to add the received soft information to the received posterior probability information after taking the negative, and output the addition result to the hierarchical decoding DEC unit and the second adder respectively; The hierarchical decoding DEC unit, the output end of the hierarchical decoding DEC unit is respectively connected to the input end of the posterior probability information storage unit and the input end of the second adder, the hierarchical decoding DEC unit is used to perform a decoding operation on the received information to obtain updated posterior probability information, and output the posterior probability information to the posterior probability information storage unit and the second adder; The a posteriori probability information storage unit is used to receive and store the a posteriori probability information, and is used to output the a posteriori probability information to the first adder when performing iterative decoding, wherein when performing the first iterative decoding, all elements in the a posteriori probability information output by the a posteriori probability information storage unit to the first adder are initialized to 0; The second adder is used for adding the received addition result to the received posterior probability information after negating it to obtain updated soft information, and outputting the soft information to the decoding soft information storage unit; The decoding decision unit is used to determine whether the set iteration stop criterion is met according to the received soft information, and if so, output the decoding result; if not, enable the LDPC code soft decision decoding device to continue iterative decoding based on the current iterative decoding result.
2. The LDPC code soft decision decoding device according to claim 1, characterized in that: The decoding soft information storage unit comprises: a first data input selection circuit, a first dual-port memory and a first data output selection circuit; The first data input selection circuit is provided with a multiplexer, a plurality of first input branches, a plurality of second input branches and a plurality of output branches, one end of the plurality of first input branches is respectively connected to the output end of the multiplexer, the input end of the multiplexer is used as the input end of the decoding soft information storage unit, and is used to receive the soft information input into the LDPC code soft decision decoding device, one end of the plurality of second input branches is used to be connected to the second adder, and the other end of the first input branch and the other end of the second input branch are switchably connected to one end of the output branch through a single-pole double-throw switch; The first dual-port memory includes a plurality of first dual-port memories, the plurality of first dual-port memories are respectively connected to the other ends of the plurality of output branches of the first data input selection circuit, and each of the first dual-port memories is configured with a read address generator and a write address generator; The first data output selection circuit is provided with multiple input ports and multiple output ports, the inner ends of the multiple input ports are respectively connected to the inner ends of the multiple output ports through control switches, the outer ends of the multiple input ports are respectively connected to the multiple first dual-port memories, and the outer ends of the multiple output ports are used to connect to the first adder and the decoding decision unit.
3. The LDPC code soft decision decoding device according to claim 2, characterized in that: Set the layered decoding DEC unit to receive information Q m,n , for receiving information Q m,n The decoding operation includes the following: Calculate information Q m,n The absolute value of Determine the absolute value |Q m,n |The minimum value min1 and the second minimum value min2 of each column, and record the row index corresponding to the minimum value min1 of each column; The absolute value |Q m,n |The elements of each column are filled with the minimum value min1 of the column and the element with the row number Index of the column is replaced with the second minimum value min2 of the column, and the matrix V is obtained. m,n ; Determine and record the positive and negative of each element in the information Qm,n, and obtain the symbol matrix sign(Q m ,n ); The symbol matrix sign(Q m,n ) multiply each column element by one, and add the multiplication result to the symbol matrix sign(Q m,n ) Multiply the elements of each column separately to get the new symbol matrix C m,n ; According to R m,n =S·V m,n ·C m,n Calculate the posterior probability information R m,n And output, where S represents the preset normalization coefficient.
4. The LDPC code soft decision decoding device according to claim 3, characterized in that: The hierarchical decoding DEC unit includes: an absolute value calculation module, a symbol determination module, a data symbol operation module, a data sorting module, a data filling module, a data replacement module and a data symbol recovery module; The absolute value calculation modules include multiple modules, the symbol determination modules include multiple modules, the input ends of the multiple absolute value calculation modules are respectively connected to the input ends of the multiple symbol determination modules and serve as the input end of the hierarchical decoding DEC unit, the output ends of the multiple absolute value calculation modules are respectively connected to the input end of the data sorting module, the output ends of the multiple symbol determination modules are respectively connected to the input end of the data symbol operation module, the input end of the data filling module is connected to the input end of the data sorting module, the output end of the data sorting module is connected to the input end of the data replacement module, the output end of the data replacement module and the output end of the data symbol operation module are respectively connected to the input end of the data symbol recovery module, and the output end of the data symbol recovery module serves as the output end of the hierarchical decoding DEC unit; The plurality of absolute value calculation modules and the plurality of sign determination modules are used to receive in parallel in the same clock cycle a plurality of data of the same column of information input into the hierarchical decoding DEC unit, and respectively calculate and output the absolute values and positive and negative signs of the plurality of data; The data symbol operation module is used to receive the positive and negative symbols input by the plurality of symbol determination modules, multiply the positive and negative symbols, and multiply the multiplication result by the positive and negative symbols respectively to obtain and output a column of elements of the symbol matrix; The data filling module is used to generate multiple data serial numbers and sequentially fill them into multiple absolute values output by multiple absolute value calculation modules; The data sorting module is used to sort the received multiple absolute values, determine the minimum value among the multiple absolute values and its corresponding data sequence number and the second smallest value among the multiple absolute values and its corresponding data sequence number, and output the received multiple absolute values, the determined minimum value and its corresponding data sequence number and the second smallest value and its corresponding data sequence number; The data replacement module is used to replace the minimum value of multiple absolute values with the next smallest value and output it, and replace other values of the multiple absolute values with the minimum value and output it according to the data sequence number corresponding to the minimum value and the data sequence number corresponding to the next smallest value; The data symbol recovery module is used to multiply each column element received from the data symbol operation module, each column element received from the data replacement module and a preset normalization coefficient to obtain and output posterior probability information.
5. The LDPC code soft decision decoding device according to claim 4, characterized in that: The posterior probability information storage unit comprises: a second data input selection circuit, a second dual-port memory, and a second data output selection circuit; The second data input selection circuit is provided with a plurality of input ports and a plurality of output ports, the inner ends of the plurality of input ports are respectively connected to the inner ends of the plurality of output ports through control switches, and the outer ends of the plurality of input ports are used to be connected to the hierarchical decoding DEC unit; The second dual-port memory includes a plurality of second dual-port memories, the plurality of second dual-port memories are respectively connected to the external ends of the plurality of output ports of the second data input selection circuit, each of the second dual-port memories is configured with a read address generator and a write address generator, and the read address generator and the write address generator configured in each of the second dual-port memories are connected to an address offset controller, and the address offset controller is capable of generating parameters for address generation; The second data output selection circuit is provided with a plurality of input ports and a plurality of output ports, the inner ends of the plurality of input ports of the second data output selection circuit are respectively connected to the inner ends of the plurality of output ports of the second data output selection circuit through control switches, the outer ends of the plurality of input ports of the second data output selection circuit are respectively connected to a plurality of the second dual-port memories, and the outer ends of the plurality of output ports of the second data output selection circuit are used to be connected to the first adder.
6. The LDPC code soft decision decoding device according to claim 5, characterized in that: The decoding decision unit includes: a plurality of register groups, a decoding verification module and a verification decision module; The multiple register groups are respectively connected to the decoding soft information storage unit and the decoding verification module, each register in the multiple register groups is used to store single-bit decoding hard decision data, the decoding verification module has multiple output ports, each output port of the decoding verification module corresponds to a row of the LDPC code check matrix, the multiple output ports of the decoding verification module are respectively connected to the input end of the check decision module, the output end of the check decision module is connected to the multiple register groups, the check decision module is used to judge the results output by the multiple output ports of the decoding verification module, and select the first multiple bits cached in the register group as the decoding result output when the judgment is passed, and enable the LDPC code soft decision decoding device to continue the next round of iterative decoding based on the current iterative decoding result when the judgment is not passed.
7. The LDPC code soft decision decoding device according to claim 6, characterized in that: The device further comprises: a control unit, the control unit being used to store decoding control parameters and perform corresponding control according to the stored decoding control parameters; The decoding control parameters include: control signals of the decoding soft information storage unit and the posterior probability information storage unit for the data input selection circuit and the data output selection circuit in the hierarchical decoding stage, initialization addresses of the read address generator and the write address generator of the decoding soft information storage unit and the posterior probability information storage unit in the hierarchical decoding stage, and address offset control amount of the posterior probability information storage unit in the hierarchical decoding stage.
