CN101777926B - General decoder of Turbo product code and method thereof - Google Patents

Abstract

The invention discloses a universal decoder for Turbo product codes and a method thereof. The universal decoder consists of an initial information storage module, an external information storage module, an unreliable bit calculation module, an algebraic decoding module, two first-in-first-out storage modules, a measurement comparison module, an external information calculation module, a decoding control module and an interface. composition. Its decoding parameters can be freely configured, that is, the two-dimensional subcode of the Turbo product code, and the number of decoding times can be selected arbitrarily, which can meet the needs of various communication systems for different code lengths, code rates, decoding delays and throughputs; The structural design of the device is flexible and can be realized by both programmable logic devices and special chips; the device adopts a pipeline structure in many places, which improves the throughput of the decoder. The invention takes into account the three indexes of universality, decoding performance and complexity, and can be conveniently and flexibly used in communication systems with various code lengths, code rates, realization platforms and bit error rate requirements.

Description

A Universal Decoder for Turbo Product Codes and Its Method

technical field

The invention relates to the technical field of mobile communication, in particular to a universal decoder for Turbo product codes and a method thereof.

Background technique

As a new type of error-correcting code, Turbo code has drawn much attention in the communication field, and its error-correcting capability can approach the Shannon limit. Since Turbo code was proposed by C.Berrou in 1993, it has become a hot spot in channel coding research. Its implementation methods are roughly divided into two types: enhanced Turbo Product Codes (TPC) and enhanced convolutional codes (Turbo Convolutional Codes). TCC). TPC shows attractive application prospects in many aspects. On the one hand, the encoding efficiency of TPC is higher than that of TCC. In addition, the main advantage of TPC is that it uses matrix interleaving, which makes the system structure relatively simple. The serial connection structure of Turbo product encoding also has many advantages over the parallel connection structure of Turbo convolutional encoding. This paper mainly studies a kind of universal Turbo Product Code (TPC) decoding. Several hard-decision and soft-decision decoding algorithms have been born during the development of product codes, which are different in implementation complexity and error correction performance. However, the Turbo product code adopts an iterative soft-input soft-output (SISO, Soft-Input Soft-Output) decoding method, and this decoding method has better performance.

However, in the face of different business requirements and communication systems, it will be inconvenient to implement a fixed single decoder. However, there is no such a universal decoder at present, and most implementations are for specific codewords or codes. Rate. Therefore, the inventor proposed this general turbo product code decoder, which can decode communication systems with different code lengths, code rates, platforms, decoding delays and throughput requirements.

Contents of the invention

The purpose of the present invention is to overcome the deficiencies in the prior art, a kind of universal decoder of Turbo product code and method thereof are provided

The configuration parameter of the universal decoder of Turbo product code has: the subcode pattern of Turbo product code, the number of iterations, and it at least includes:

The initial information storage module is used to store the receiving sequence required for each decoding;

The external information storage module is used for storing the external soft information required and obtained for each decoding;

The first first-in first-out module or the second first-in first-out module is used to temporarily store the decoded input sequence;

The unreliable bit calculation module converts the input sequence according to the unreliable bit in the external configuration parameters to obtain the input sequence of algebraic decoding;

The algebraic decoding module performs different algebraic decoding according to different configuration parameters;

The measurement comparison module obtains the optimal codeword D by comparing the Euclidean distance of the output of the algebraic decoding module;

The external information calculation module calculates the external soft information of this iteration and provides it to the next iteration;

The control module controls the timing and parameter selection of the above-mentioned various modules;

Interface module for parameter setting;

The external information storage calculation module is connected with the input of the first first-in first-out module, the least reliable position calculation module and the external information module respectively; the initial information storage module is connected with the first first-in-first-out module and the least reliable position calculation module respectively; The first first-in-first-out module is connected with the external information storage calculation module, the initial information storage module, the second first-in-first-out module, and the measurement comparison module respectively; the least reliable bit calculation module is respectively connected with the initial information storage module, the external information storage module, The algebraic decoding module is connected; the algebraic decoding module is connected with the least reliable calculation module and the measurement comparison module respectively; the second first-in-first-out module is respectively connected with the first first-in-first-out module and the external information module input; the measurement comparison module They are respectively connected with the input of the first first-in-first-out module, algebraic decoding module, and external information module; the input of the external information module is respectively connected with the measurement comparison module, the second first-in-first-out module, and the external information storage and calculation module; the above-mentioned modules are controlled by Module control; parameterization in the interface module.

The subcode pattern of the Turbo product code can be any Hamming code with the longest code length of 64 or extended Hamming code and its shortened code, and the number of shortened bits is any number, and the maximum value is the length of the code word.

The number of iterations is an arbitrary number set according to decoding delay and bit error rate.

In the external information storage module, the capacity of the initial information storage module is 64 units*64 units, and the capacity of each unit is the number of quantized bits.

The capacity of the first first-in-first-out module or the second first-in-first-out module is 2*64 units, and the capacity of each unit is the quantized bit of decoding input soft information, and the readout order is the order in which data is written , the data in the FIFO module is empty after the data is read.

The algebraic decoding module includes arbitrary Hamming codes or extended Hamming codes and shortened codes with component code lengths up to 64.

The general decoding method of Turbo product code comprises the following steps:

1) Parameterized configuration, set the two-dimensional subcode and the number of iterations of Turbo product code decoding in the interface module as required;

2) receiving the demodulation symbol information output by the channel, storing the received information sequence required for this decoding in the initial information storage module, the initial information of the external information storage module is 0, and starting an iterative decoding of a row or column;

3) Calculate the sequence information required for decoding and store it in a first-in-first-out module, and at the same time find the least reliable 3 bits, that is, the 3 bits with the smallest measurement value, and take these 3 codewords to be 0 or 1 respectively, and get 8 candidate codeword sequences;

4) Algebraically decoding these 8 codewords in the algebraic decoding module, the algebraic decoding methods of different configurations are different, and obtain 8 codewords after algebraic decoding;

5) Send the output codeword of algebraic decoding and the output of the first first-in-first-out module to the measurement comparison module, find the codeword with the smallest Euclidean distance, and determine it as the optimal codeword for the row or column, and at the same time the second The block FIFO module reads in the output of the first block FIFO module;

6) Send the output of the second first-in-first-out module and the output of the metric comparison module to the external information calculation module, calculate the external soft information of this row or column iteration, and store it in the external soft information memory. Or the decoding of the column is completed;

7) Repeat the above steps until all rows or columns are decoded, then perform column or row decoding, and one iteration ends after the whole is completed;

8) Repeat the above steps, according to the parameter configuration, the control module controls the whole process until the whole iteration ends;

The decoding between row and row, column and column of Turbo product code subcode adopts pipeline structure.

The present invention has realized code length, code rate and number of iterations are variable, and multi-platform streamlined Turbo product code decoding, and the decoding performance of the decoder for Turbo product codes composed of different subcodes and different iteration times A series of tests were performed. This universal decoder has strong portability and can be applied to different communication systems.

Description of drawings

Fig. 1 is the schematic diagram of the general decoder of Turbo product code of the present invention;

Fig. 2 is a general block diagram of internal structure of Turbo product code decoder of the present invention;

Fig. 3 is each decoding process of the general decoder of Turbo product code of the present invention;

Fig. 4 is the structure of the first-in-first-out module of the present invention;

Fig. 5 is the workflow of algebraic decoder of the present invention;

Fig. 6 is the pipeline structure of the measurement comparison module of the present invention;

Fig. 7 is the iterative decoding process of the Turbo product code universal decoder of the present invention;

Fig. 8 is (15,11)*(15,11) Hamming code and (31,26)*(31,26) the Turbo product code decoding performance test graph that Hamming code constitutes;

Fig. 9 is (31,26)*(31,26) Hamming code and (32,26)*(32,26) extended Hamming code decoding performance test curve diagram of the Turbo product code;

Fig. 10 is a graph showing decoding performance test curves of Turbo product codes with different iteration times. The present invention will be further described below in conjunction with accompanying drawing:

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be further described below in conjunction with the accompanying drawings:

The universal decoder of Turbo product codes can decode communication systems with different code lengths, code rates, platforms, decoding delays and throughput requirements. The two-dimensional subcode can be selected as any Hamming code with the longest code length of 64 or the extended Hamming code and its shortened code, and the number of shortened digits is any number, and the maximum value is the length of the code word. In this way, it can be used for different code lengths and code rates; the number of decoding times can be selected arbitrarily to meet the requirements of decoding delay and throughput; its structured design is flexible and scalable, and can be realized by programmable logic devices , such as FPGA, can also be implemented with a dedicated chip (ASIC); multiple pipeline structures are used to fully improve the throughput of the decoder. The utility model can be applied to different communication systems, and is easy to use and has a simple and clear structure.

The general decoder of Turbo product code, its configuration parameter has: the subcode pattern of Turbo product code, the number of iterations, the internal structure of this Turbo product code general decoder is as shown in Figure 2, and it at least includes:

An initial information storage module 202, configured to store the received sequence required for each decoding;

The external information storage module 201 is used for storing the external soft information required and obtained for each decoding;

The first first-in first-out module 203 or the second first-in first-out module 206 is used for temporarily storing the decoded input sequence;

The unreliable bit calculation module 204 converts the input sequence according to the unreliable bit in the external configuration parameters, and obtains the input sequence of algebraic decoding;

The algebraic decoding module 205 performs different algebraic decoding according to different configuration parameters;

The metric comparison module 207 obtains the optimal codeword D by comparing the Euclidean distance of the output of the algebraic decoding module;

The external information calculation module 208 calculates the external soft information of this iteration and provides it to the next iteration;

The control module 209 controls the timing and parameter selection of the above-mentioned various modules;

Interface module 101, for parameter setting;

The external information storage calculation module 201 is connected with the first FIFO module 203, the least reliable bit calculation module 204, and the external information module input 208 respectively; the initial information storage module 202 is respectively connected with the first FIFO module 203, the least reliable bit The bit calculation module 204 is connected; the first first-in-first-out module 203 is respectively connected with the external information storage calculation module 201, the initial information storage module 202, the second first-in-first-out module 206, and the measurement comparison module 207; the least reliable bit calculation module 204 is connected with initial information storage module 202, external information storage module 201, algebraic decoding module 205 respectively; Algebraic decoding module 205 is connected with least reliable calculation module 204, metric comparison module 207 respectively; The second first-in-first-out module 206 is connected with first FIFO module 203, external information module input 208 respectively; Metric comparison module 207 is connected with first FIFO module 203, algebraic decoding module 205, external information module input 208 respectively; External information module The input 208 is respectively connected with the metric comparison module 207 , the second FIFO module 206 , and the external information storage and calculation module 201 ; the above modules are controlled by the control module 209 ; parameters are set in the interface module 101 .

The subcode pattern of the Turbo product code can be any Hamming code with the longest code length of 64 or extended Hamming code and its shortened code, and the number of shortened bits is any number, and the maximum value is the length of the code word.

The number of iterations is an arbitrary number set according to decoding delay and bit error rate.

The external information storage module 201 and the initial information storage module 202 have a capacity of 64 units*64 units, and the capacity of each unit is its quantized bit number.

The capacity of the first first-in-first-out module 203 or the second first-in first-out module 206 is 2*64 units, and the capacity of each unit is the quantized bit of decoding input soft information, and the readout sequence is data writing After the data is read out, the data in the FIFO module is empty.

The algebraic decoding module 205 includes any Hamming code or extended Hamming code and its shortened code with a component code length of up to 64.

The schematic diagram of a general-purpose Turbo decoder is shown in Figure 1. The optional parameter of the Turbo two-dimensional product code to be decoded is the number of iterations P1.1, which can be set according to the complexity requirements; the horizontal and vertical subcode types Parameter P1.2, which includes any Hamming code or extended Hamming code and its shortened code with the longest code length of 64, and the number of shortened digits is any number, and the maximum value is the length of the code word.

The internal parameter configurations of the turbo product code decoders of these different formats are different, so the set parameters are passed into the interior through the interface module to realize the universal turbo product code decoding.

The general decoding method of Turbo product code comprises the following steps:

1) Parameterized configuration, set the two-dimensional subcode and the number of iterations of Turbo product code decoding in the interface module as required;

2) receiving the demodulation symbol information output by the channel, storing the received information sequence required for this decoding in the initial information storage module, the initial information of the external information storage module is 0, and starting an iterative decoding of a row or column;

3) Calculate the sequence information required for decoding and store it in a first-in-first-out module, and at the same time find the least reliable 3 bits, that is, the 3 bits with the smallest measurement value, and take these 3 codewords to be 0 or 1 respectively, and get 8 candidate codeword sequences;

4) Algebraically decoding these 8 codewords in the algebraic decoding module, the algebraic decoding methods of different configurations are different, and obtain 8 codewords after algebraic decoding;

5) Send the output codeword of algebraic decoding and the output of the first first-in-first-out module to the measurement comparison module, find the codeword with the smallest Euclidean distance, and determine it as the optimal codeword for the row or column, and at the same time the second The block FIFO module reads in the output of the first block FIFO module;

6) Send the output of the second first-in-first-out module and the output of the metric comparison module to the external information calculation module, calculate the external soft information of this row or column iteration, and store it in the external soft information memory. Or the decoding of the column is completed;

7) Repeat the above steps until all rows or columns are decoded, then perform column or row decoding, and one iteration ends after the whole is completed;

8) Repeat the above steps, according to the parameter configuration, the control module controls the whole process until the whole iteration ends;

The specific decoding process of each iteration is shown in Figure 3.

First, the above step 2 is to read the received sequence and soft information from the initial information storage module 202 and the external information storage module 201, because the maximum length of the configurable codeword is 64, so the size of the RAM is 64 units×64 unit, the size of each unit is the number of quantized bits of extrinsic information and soft information, and the soft input is calculated through the received information, for the mth iteration

R(m)＝R+α(m)·W(m)

Among them, R - the signal received by the decoder

R(m)——the soft input signal of the mth unit decoder

W(m)——Extrinsic information output by the upper-level unit decoder, the initial value is zero

α(m)——Adjustment coefficient

Since the sampling standard deviations of [R] and [W(m)] are different, the sampling standard deviation of [W(m)] is relatively large at the beginning, and gradually decreases as the number of iterations increases, so it is necessary to introduce an adjustment coefficient α (m), α(m) plays a role in suppressing [W(m)] when the signal-to-noise ratio is large at the initial stage of decoding. The value of α(m) increases with the iterative process. In the later stage of decoding, the influence of the initial input signal [R] on the soft input gradually decreases, and the reliability of the value of W continues to increase, which can be used as the main factor for decoding. According to, its influence on the soft input gradually increases to achieve the best decoding effect. After the soft input value is obtained, soft decision decoding is performed on the input value, and the soft input information is decoded. Simultaneously, soft information is stored in the first-in-first-out module 203, and the structure of the first-in-first-out module is as shown in Figure 4, and when there is a write enable signal, R (m) is stored in the first-in first-out module; When the signal is enabled, R(m) reads out the first-in-first-out module; the first-in-first-out module is empty after reading out all the modules. To prevent overflow, set its size to 64 units x 2.

The second step, that is, the above step 3 is to perform a hard decision on the soft input sequence r in the least reliable bit calculation module 204 to obtain the sequence z, and assign a reliability value to each bit of z; take the p bit with the lowest reliability in z , take these p positions to be 0 or 1 respectively to get 2 ^p error patterns, and add these error patterns to the sequence z obtained by hard judgment to generate a candidate codeword set Ω;

In an embodiment, let p = 3, there are at most 8 different codewords in Ω. In this way, a compromise can be made between decoding performance and complexity, which is easier to implement.

The third step, that is, the above step 4 is to perform algebraic decoding on each codeword in the candidate codeword set Ω to obtain the codeword v. For Turbo product codes with different parameters, different algebraic decoding is used, which is universal The core, as shown in Figure 5. In the embodiment, the code word of (15,11)×(15,11) and (31,26)×(31,26) are selected, and the Hamming code word with or without check digit is tested;

For the Hamming codeword without check digit, directly carry out algebraic decoding as embodiment (15,11)×(15,11) and (31,26)×(31,26), for the Hamming code word with check digit Words are first algebraically decoded, and then processed differently according to the value of the error mode err and the parity check bit c.

For the shortened Hamming code, fill in 0 first, assign a high confidence value to the 0-fill bit, search for the least reliable bit, and then perform algebraic decoding as above.

At the same time, the information in the first first-in first-out module 203 is input into the second first-in first-out module 206 .

The fourth step, that is, the above-mentioned step 5 is to calculate the soft-decision decoding metric of each candidate codeword in the metric comparison module 207, and select the most probable codeword. Usually, the Euclidean distance between each candidate codeword and the soft input sequence is calculated, and the codeword with the smallest Euclidean distance is selected as the decision codeword D.

The fifth step, that is, the above step 3 is to calculate the competing codeword C, and each bit of the decision codeword D needs to be searched. Therefore, the information of the second first-in-first-out module 206 is input into the external information calculation module 208, and in the external information calculation module 208, each bit in D is searched in the 8 code words of the set Ω to see if there is a corresponding position If there are codewords with opposite signs, among these codewords with opposite signs, the codeword with the smallest Euclidean distance to the soft input vector of this decoding step is the competing codeword. If there is no codeword of opposite sign, then there is no competing codeword for this bit. The specific embodiment is shown in Figure 6, which adopts a flow structure. XOR the i-th bit data[i] of the 8 candidate codewords with the optimal codeword D, if it is 0, assign its corresponding Euclidean distance to 0; if it is 1, keep its Euclidean distance dis , and then get four relatively large distance values dismin1, dismin2, dismin3, dismin4 by pairwise comparison, and then compare these four codewords pairwise to obtain relatively large distance dismaxa, dismaxb, and then compare these two values to get the maximum The codeword corresponding to the Euclidean distance, that is, the competing codeword C.

If a competing codeword can be found, the calculation method of the external information of this bit is as follows:

${\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; [</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 4</span>}}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}$

Among them, R——the soft input of the decoder of this stage

C - the competing codeword for this bit

D——Decision codeword obtained by soft-decision decoding

dj - the jth bit of the decision codeword D

j - the jth bit of the soft input R

It can be seen from the above formula that the sign of the soft output r _j =r _j +w _j is the same as that of dj, and its absolute value represents the reliability of the soft output; if no competing code word is found in the candidate code word set Ω, it means The Euclidean distance between C and R is very large, that is, the reliability of dj is very high; if the Euclidean distance between C and R is very long, it means that the reliability of dj is very high,

If no competing codeword is found, the calculation method is:

w _j =β·dj

It can be seen from the above formula that to calculate the extrinsic information, it is also necessary to calculate the value of β. β(m) is used to control the output peak range of soft output information, it can be determined through experiments, and can also be estimated by some methods. The value of B can be initialized by experimentation and error correction. Simplified algorithms are often used. β can be estimated by the reliability value of the soft input, and the reliability value of each bit in the decision codeword D without competing codewords is averaged. The β value obtained in this way is less accurate, but it is easy to implement.

After the value of the extrinsic information is calculated, it is stored in the extrinsic information storage module 201, and then the soft output can be further calculated. The value of the soft output is equal to the value of the soft input plus the external information, namely

r'=r+w

A complete iteration is divided into one row decoding and one column decoding. The specific decoding steps are the same. The alternate decoding of rows and columns constitutes the iterative decoding process of BPTC, as shown in Figure 7. Depending on the parameter settings, different number of iterations are performed. The resulting decoded codeword is finally output. In addition, the decoding between row and row, and column and column of the Turbo product code subcode adopts a pipeline structure, which increases the decoding speed and improves the throughput.

As shown in FIGS. 8, 9, and 10, the bit error rate curves of the above-mentioned specific implementation schemes of the present invention under the condition of random bit interference (AWGN channel) are given. The simulation conditions used are BPSK modulation, α{0.0625, 0.125, 0.1875, 0.25, 0.3125, 0.375, 0.4375, 0.5, 0.5625, 0.625, 0.6875, 0.75, 0.8125, 0.875, 0.9375, 1}. Figure 8 shows the decoding performance of two TPC codes (15,11)*(15,11) and (31,26)*(31,26) for Turbo product codes composed of subcodes of different lengths. Figure 9 shows the decoding performance of the TPC code with or without adding check digits. Figure 10 is the performance comparison when the number of iterations is different. As the number of iterations increases further, the improvement of decoding performance gradually slows down; when the number of iterations increases to a certain extent, the performance remains basically unchanged and can no longer be improved. Therefore, when decoding, we need to select an appropriate number of iterations to ensure low complexity while achieving optimal performance.

As mentioned above, the greatest feature of the present invention is that it can decode Turbo product codes composed of different subcodes. Taking speed and flexibility into consideration, this embodiment selects FPGA to test this general Turbo product code decoder. However, the structural design of the device can also be realized with an application-specific chip (ASIC), and in the case of a small data throughput, it can also be realized with a digital signal processor (DSP). In addition, combined with the general design process of computer software, software can be used to Implementation is also possible.

The invention is a universal turbo product code decoder and its method. The methods and circuits described here, the individual parts separated from each other can be completely conventional, and we claim their combination as an invention. The above is only a specific embodiment of a specific application, but the true spirit and scope of the present invention are not limited thereto. Any person skilled in the art can modify the specific method of a single component to realize Turbo product codes in different applications decoding. The present invention is limited only by the appended claims and their equivalents, which we claim to be protected as the present invention.

Claims (9)

1. a general decoder for Turbo product code, its configuration parameter has: the subcode pattern of Turbo product code, iterations, is characterized in which comprises at least:

Initial information memory module (202), the receiving sequence needing for storing each decoding;

External information memory module (201), for storing each decoding soft information in outside that need and that obtain;

The first first-in first-out module (203) and the second first-in first-out module (206), for the temporary transient list entries that stores decoding;

Least reliable bits computing module (204), converts list entries according to the unreliable figure place in exterior arrangement parameter, obtains the list entries of algebraic decoding;

Algebraic decoding module (205), carries out different algebraic decodings according to configuration parameter difference;

Tolerance comparison module (207), by comparing the Euclidean distance of the output of algebraic decoding module, obtains optimum code word D;

External information computing module (208), calculates the soft information in outside of this iteration, offers next iteration;

Control module (209), the sequential to above-mentioned various modules, parameter is selected to control;

Interface module (101), carries out parameter setting;

External information memory module (201) is respectively with the first first-in first-out module (203), least reliable bits computing module (204), external information computing module (208) are connected; Initial information memory module (202) is respectively with the first first-in first-out module (203), least reliable bits computing module (204) is connected; The first first-in first-out module (203) is connected with external information memory module (201), initial information memory module (202), the second first-in first-out module (206), tolerance comparison module (207) respectively; Least reliable bits computing module (204) is connected with initial information memory module (202), external information memory module (201), algebraic decoding module (205) respectively; Reliable bits computing module (204), tolerance comparison module (207) are connected algebraic decoding module (205) respectively with least; The second first-in first-out module (206) is connected with the first first-in first-out module (203), external information computing module (208) respectively; Tolerance comparison module (207) is connected with the first first-in first-out module (203), algebraic decoding module (205), external information computing module (208) respectively; External information computing module (208) is connected with tolerance comparison module (207), the second first-in first-out module (206), external information memory module (201) respectively; External information memory module (201) to external information computing module (208) is controlled by control module (209); In interface module (101), carry out parameter setting.

2. the general decoder of a kind of Turbo product code according to claim 1, is characterized in that, the subcode pattern of described Turbo product code is that maximum length code length is any Hamming code of 64.

3. the general decoder of a kind of Turbo product code according to claim 1, is characterized in that, the subcode pattern of described Turbo product code is extended hamming code.

4. the general decoder of a kind of Turbo product code according to claim 1, is characterized in that, the shortening code that the subcode pattern of described Turbo product code is Hamming code, and to shorten figure place be Arbitrary Digit, the code length that maximum is subcode.

5. the general decoder of a kind of Turbo product code according to claim 1, is characterized in that, described iterations is according to the Arbitrary Digit of decoding delay and error rate setting.

6. the general decoder of a kind of Turbo product code according to claim 1, it is characterized in that, described external information memory module (201), the capacity of initial information memory module (202) is 64 unit * Unit 64, and the capacity of each unit is in unit, to store the quantizing bit number of information.

7. the general decoder of a kind of Turbo product code according to claim 1, it is characterized in that, the first described first-in first-out module (203) and the capacity of the second first-in first-out module (206) are 2*64 unit, the capacity of each unit is the quantization bit that soft information is inputted in decoding, wherein to read order be the order that data write, after data reading, data be sky in first-in first-out module.

8. use a general decoding method for the Turbo product code of general decoder as claimed in claim 1, it is characterized in that comprising the following steps:

1) parametrization configures, and sets as required two-dimentional subcode and the iterations of Turbo product code decoding in interface module;

2) the demodulation symbol information of receive channel output, deposits the required received information sequence of decoding in initial information memory module in, and external information memory module initial information is 0, starts once the iterative decoding of row or row;

3) calculate the required sequence information of decoding, deposit in the first first-in first-out module, find least reliable 3 is 3 of metric minimum simultaneously, and this 3 bit word is got respectively to 0 or 1, obtains 8 candidate codewords sequences;

4) these 8 candidate codewords sequences are carried out to algebraic decoding in algebraic decoding module, the algebraic decoding method of different configurations is different, obtains the code word after 8 algebraic decodings;

5) output of the output codons of algebraic decoding module and the first first-in first-out module is sent into tolerance comparison module, find the code word of Euclidean distance minimum, be judged to the optimum code word of the capable or row of this iterative decoding, the second first-in first-out module is read in the output of the first first-in first-out module simultaneously;

6) output of the output of the second first-in first-out module and tolerance comparison module is sent into external information computing module, calculate the soft information in outside of this iterative decoding row or column iteration, and deposit in external information memory module, this iterative decoding is capable or column decoding is complete;

7) repeat above-mentioned steps, until all row or column decoding are complete, be listed as or row decoding, whole after an iteration finish;

8) repeat above-mentioned steps, according to parameter configuration, control module is controlled whole process until whole iteration finishes.

9. the general decoding method of a kind of Turbo product code according to claim 8, is characterized in that, row and the row of described Turbo product code subcode, and the decoding between row and row adopts pipeline organization.

Images