CN101964665A - Log-MAP based decoding method and decoding device thereof in turbo decoding - Google Patents
Log-MAP based decoding method and decoding device thereof in turbo decoding Download PDFInfo
- Publication number
- CN101964665A CN101964665A CN 201010516875 CN201010516875A CN101964665A CN 101964665 A CN101964665 A CN 101964665A CN 201010516875 CN201010516875 CN 201010516875 CN 201010516875 A CN201010516875 A CN 201010516875A CN 101964665 A CN101964665 A CN 101964665A
- Authority
- CN
- China
- Prior art keywords
- data
- input
- output
- logsum
- recursion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Error Detection And Correction (AREA)
Abstract
The invention relates to a method and a device thereof in the technical field of mobile communication technologies, in particular to a Log-MAP based decoding method and a decoding device thereof in turbo decoding. In the invention, in a Log-MAP decoding structure, the realizations of a forward state measurement and recursion structure and a backward state measurement and recursion structure are separated from the realization of an LLR computation structure; in LLR computation, an eight-input Logsum structure is not realized by adopting a two-input Logsum structure any more; the recursion of forward state measurement and the backward state measurement is realized by adopting a Max-Log-Map structure, so that the two-input Logsum structure still maintains a lowest realization complexity, meanwhile, the realization complexity of the eight-input Logsum structure is also reduced; and most importantly, the decoding properties and good properties close to an optimal Log-Map method are also maintained. The invention obtains good balance between the realization complexity and the properties of a decoder.
Description
Technical field
What the present invention relates to is the method and the device thereof in a kind of mobile communication technology field, specifically is based on interpretation method and the code translator thereof of Log-MAP in the turbo decoding.
Background technology
Turbo code since proposing in 1993, because its excellent properties near shannon limit, be widely used in the mobile communication technology field, be determined one of channel decoding scheme that is used for 3-G (Generation Three mobile communication system) and LTE (Long Term Evolution) mobile communication system.
The coding/decoding method of Turbo employing at present mainly is MAP (maximum a posteriori, maximum a posteriori probability) class methods, generally in the enterprising row operation of log-domain, all multiplyings all are converted to addition and the computing of tabling look-up, therefore helping hardware decoders realizes, reduce complexity, be referred to as the Log-Map method.But the hardware implementation complexity of Log-Map is still very high, main complexity is forward state metric recursion structure in the decoding architecture, the back is to the realization of state measurement recursion structure and LLR (bit log-likelihood ratio) computation structure, above structure is commonly referred to the Logsum structure, in a Log-Map decoder, generally include three Logsum structures, a recursion that is used for forward state metric, a recursion that is used for the back to state measurement, also has a calculating that is used for LLR, wherein being used for forward state metric and back is two parameters to the input interface of the Logsum of state measurement recursion structure, be referred to as two input Logsum structures, this structure mainly is to be achieved as follows computing: and Logsum (x, y)=ln (e
x+ e
y); The input interface that is used for the Logsum structure of LLR calculating is eight parameters, is referred to as eight input Logsum structures, and this structure mainly is to be achieved as follows computing:
But it is to realize that with two input Logsum structures essence is the same.The Logsum structure will realize the exponential sum logarithm operation, if the realization of Logsum structure is too simple, and then decoded decreased performance, then complexity is too high again to keep performance, and therefore, the compromise between performance and the complexity considers it is the key that realizes.
Existing technology all is to realize eight input Logsum on the basis of realizing two input Logsum structures, the realization of two input Logsum has two kinds of structures, first kind is that two input Logsum structures are only got two maximums in the input, this method is referred to as Max-Log-Map two input Logsum structures, this method has reduced complexity, but performance is also lost maximum; To be two input Logsum add a correction again getting on the peaked basis for second kind of structure, usually use the interpolation of look-up tables'implementation correction, the calculating of correction has several different methods, and this method has kept performance near break-even Log-Map, but complexity is but very high.
More than two kinds of structures be with the forward state metric recursion, the structure of eight input Logsum bound together during the structure of back two input Logsum in the state measurement recursion and LLR calculated, thereby the result who causes is or the complexity height, poor performance, between complexity and performance, do not obtain good compromise, in real system is realized, because the Turbo complexity of decoding is very high, thereby Logsum generally adopts Max-Log-Map two input Logsum structures, and this has caused the loss of performance inevitably.
Summary of the invention
The purpose of invention is to overcome the deficiencies in the prior art, and interpretation method and code translator thereof based on Log-MAP are provided in a kind of turbo decoding.Make the complexity of deciphering based on Log-MAP in its Turbo decoding low, simultaneously functional.
The present invention is achieved by the following technical solutions:
Based on the interpretation method of Log-MAP, comprise the steps: in the turbo decoding that the present invention relates to
The branch metric that step 2, usefulness calculate comes recursion to calculate forward state metric, in recursive process, need to realize two input Logsum computings, the present invention adopts the Max-Log-Map structure herein, and promptly two input Logsum are output as the maximum of two inputs, make it keep minimal complexity.
Will realize eight input Logsum computings described in the step 5 in calculating the LLR process, this new method is made up of following five small steps:
The first step is by relatively obtaining eight maximums in the input;
Second step, this maximum subtracted each other with these eight inputs respectively and judge each difference whether in the thresholding of setting, if then export one, otherwise output zero, thresholding herein is a predefined constant;
In the 3rd step, will deduct one again after these output additions;
In the 4th step, the value that obtains is searched a good as calculated look-up table as the subscript method;
The 5th step was exactly the approximation of eight input Logsum with the maximum addition result of searching in eight inputs that the value that obtains and the first step before obtain.Value by the eight input Logsum that obtain calculates LLR.
The invention still further relates to as in the above-mentioned turbo decoding based on the code translator of the interpretation method of Log-MAP, comprise: branch metric calculation unit, forward state metric recursion unit, ram cell, back are to state measurement recursion unit and LLR computing unit, the branch metric calculation unit receives the input data, is connected with LLR computing unit three to transmit to state measurement recursion unit with forward state metric recursion unit, back respectively simultaneously to calculate branch metric; Forward state metric recursion unit receives the data of branch metric calculation unit output, forward-facing state degree recursion unit is connected the data that transmit output with ram cell, and temporarily store by ram cell, ram cell is connected with the LLR computing unit that transmit will the forward state metric data of back after the state measurement recursion is calculated; The back is connected with the LLR computing unit to state measurement recursion unit, and the back goes out the back to state measurement to the data recursion that state measurement recursion unit receives the output of branch metric calculation unit; The LLR computing unit receives the branch metric calculation unit, and ram cell and back calculate LLR to the data of state measurement recursion unit output.
Described forward state metric recursion unit adopts the Max-Log-Map structure with back two input Logsum in state measurement recursion unit.
The structure of eight input Logsum comprises level Four comparator, two-stage subtracter, two stage adder, two-stage delayer and a look-up table in the described LLR computing unit, eight input data are connected with eight first order delayers with four first order comparators simultaneously, four first order comparators receive eight input data, each comparator receives two input data, and first order comparator is connected the maximum value data that transmits output with second level comparator; Two second level comparators receive four data of first order comparator output, and second level comparator is connected the maximum value data that transmits output with third level comparator; A third level comparator receives two data of second level comparator output, and third level comparator is connected the maximum value data that transmits output with eight first order subtracters respectively simultaneously with a second level delayer; A second level delayer receives the data of third level comparator output, with five clocks of this data delay, and second level delayer and data after second level adder is connected propagation delay; Eight first order delayers receive eight input data, and each delayer receives input data and postpones three clocks, and each first order delayer is connected the data that transmit output respectively with a first order subtracter; Eight first order subtracters receive the data of first order delayer and the output of third level comparator, each first order subtracter all receives the data of third level comparator output, receive the data of a first order delayer output simultaneously, first order subtracter is connected poor that transmission calculates with fourth stage comparator; Eight fourth stage comparators receive the difference and identical thresholding constant T of a first order subtracter output respectively separately, if difference greater than T, exports one, otherwise output zero, fourth stage comparator is connected the data that transmit output with first order adder; First order adder receives the data of eight fourth stage comparators outputs, calculate they and, first order adder is connected with second level subtracter that transmission calculates with; Second level subtracter receive the output of first order adder and with constant one, should and deduct one, second level subtracter is connected poor that transmission calculates with look-up table; Look-up table receives the data of second level subtracter output, is that subscript is searched with these data, and look-up table is connected the value that transmission finds with second level adder; Second level adder receives the data of look-up table output and the data of second level delayer output, and summation obtains Logsum.
The present invention has made significant improvement to prior art, the technological means of existing interpretation method is that eight input Logsum structures and two input Logsum structure connections are in the same place, how two input Logsum structures realize just having determined how eight input Logsum structures realize that its result is exactly complexity height or poor performance; The property of the present invention is directed to ground splits the realization of eight input Logsum structures and two input Logsum structures and comes, realize separately, two input Logsum structures realize with the simplest structure, eight input Logsum structures are with a kind of structure realization of new low complex degree, this fractionation realizes and utilizes eight input Logsum structures promptly can keep overall complexity very low, can keep good performance again.
Say that further separate to the realization of state measurement recursion structure and the realization of LLR computation structure forward state metric recursion structure and back in the Log-Map decoding architecture; Eight input Logsum structures no longer adopted two input Logsum structures to realize during LLR calculated, forward state metric adopts the Max-Log-Map structure to realize in the recursion of state measurement with the back, make two input Logsum structures still keep minimum implementation complexity, reduce simultaneously the implementation complexity of eight input Logsum structures again, the more important thing is the performance and the optimum good performance of Log-Map method that have also kept decoding.
The present invention has obtained tangible technological progress and outstanding beneficial effect:
Code translator provided by the invention and method have obtained fine balance between decoder implementation complexity and performance, make forward state metric recursion and back keep minimum complexity to the implementation structure of state measurement recursion, and the calculating implementation structure of LLR adopts eight input Logsum structures of the middle compared to existing technology band correction of eight input Logsum structures provided by the invention also to reduce complexity, simultaneously because eight input Logsum structures provided by the invention are all better than the break-even eight input Logsum structural behaviours that adopt index and logarithm to realize, can compensate a part of correction that lose because two input Logsum adopt the Max-Log-Map structures forward state metric and back in the state measurement recursion, thereby make whole decoding performance also keep finely, compare the Log-Map structural behaviour and can not lose too much.
Description of drawings
Fig. 1 is the decoding dress figure based on Log-MAP decoding;
Fig. 2 is Max-Log-Map two input Logsum structure charts;
Fig. 3 is two input Logsum structure charts of band correction;
Fig. 4 is the structure chart that eight input Logsum realize with two input Logsum;
Fig. 5 is eight input Logsum structure charts provided by the invention;
Fig. 6 chooses reference diagram for threshold parameter of the present invention, correction function z=lg (1+e
-| x-y|);
Fig. 7 is an embodiment error performance comparison diagram.
Embodiment
Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the present invention program, provided detailed execution mode and concrete operating process, but protection scope of the present invention is not limited to following embodiment.
Embodiment
Present embodiment relates in a kind of turbo decoding based on the interpretation method of Log-MAP, comprises the steps:
After step 4, the initialization to state measurement: B
N(1)=0, B
N(2: 8)=-∞, come recursion to calculate the back with the branch metric that calculates to state measurement, in recursive process, also need to realize two input Logsum computings, present embodiment still adopts the Max-Log-Map structure herein;
Step 9, the signal to noise ratio that changes, the above decoding of repetition step obtains the decode results under each signal to noise ratio.
Realization eight input Logsum computings described in the step 5, finish according to following five small steps:
The first step is by relatively obtaining eight maximums in the input;
Second step, this maximum subtracted each other with these eight inputs respectively and judge each difference whether in the thresholding of setting 4, if then export one, otherwise output zero;
In the 3rd step, deduct one again after the output addition with second step;
The 4th step, the value that obtains is removed to search a good as calculated look-up table as subscript, this look-up table is [00.2859 0.5718 0.8578 1.1437 1.4296 1.7155 2.0015] in the present embodiment;
The 5th step was exactly the approximation of eight input Logsum with the maximum addition result of searching in eight inputs that the value that obtains and the first step before obtain.Value by the eight input Logsum that obtain calculates LLR.
As shown in Figure 1, present embodiment also relate to as in the above-mentioned turbo decoding based on the code translator of the interpretation method of Log-MAP, comprise: branch metric calculation unit (10), forward state metric recursion unit (11), RAM (memory) unit (12), the back is to state measurement recursion unit (13), LLR computing unit (14), branch metric calculation unit (10) receives the input data, is connected with LLR computing unit (14) three to transmit to state measurement recursion unit (13) with forward state metric recursion unit (11), back respectively simultaneously to calculate branch metric; Forward state metric recursion unit (11) receives the data of branch metric calculation unit (10) output, forward state metric recursion unit (11) is connected the data that transmit output with RAM (memory) unit (12), and temporarily store by RAM (memory) unit (12), RAM (memory) unit (12) is connected with LLR computing unit (14) that transmit will the forward state metric data of back after the state measurement recursion is calculated; The back is connected with LLR computing unit (14) to state measurement recursion unit (13), and the back goes out the back to state measurement to the data recursion that state measurement recursion unit (13) receives branch metric calculation unit (10) output; LLR computing unit (14) receives branch metric calculation unit (10), and ram cell (12) and back calculate LLR to the data of state measurement recursion unit (13) output.
The described forward state metric recursion of present embodiment unit (11) adopts Max-Log-Map two input Logsum structures to realize with back two input Logsum in state measurement recursion unit (13); Max-Log-Map two input Logsum structures are two maximums in the input for the output result, as shown in Figure 2.
Eight input Logsum adopt structure provided by the invention to realize in the LLR computing unit described in the present embodiment code translator (14), as shown in Figure 5,
Eight input Logsum structures of present embodiment comprise level Four comparator, two-stage subtracter, two stage adder, two-stage retard device and a look-up table.Eight input data are connected with eight first order delayers (144) with four first order comparators (141) simultaneously, four first order comparators (141) receive eight input data, each comparator receives two input data, and first order comparator (141) is connected the maximum value data that transmits output with second level comparator (142); Two second level comparators (142) receive four data of first order comparator (141) output, and second level comparator (142) is connected the maximum value data that transmits output with third level comparator (143); A third level comparator (143) receives two data of second level comparator (142) output, and third level comparator (143) is connected the maximum value data that transmits output with eight first order subtracters (145) respectively simultaneously with a second level delayer (147); A second level delayer (147) receives the data of third level comparator (143) output, with five clocks of this data delay, and second level delayer (147) and data after second level adder (1411) is connected propagation delay; Eight first order delayers (144) receive eight input data, each first order delayer (144) receives input data and postpones three clocks, and each first order delayer (144) is connected the data that transmit output respectively with a first order subtracter (145); Eight first order subtracters (145) receive the data of first order delayer (144) and third level comparator (143) output, each first order subtracter (145) all receives the data of third level comparator output (143), receive the data of a first order delayer (144) output simultaneously, first order subtracter (145) is connected poor that transmission calculates with fourth stage comparator (146); Eight fourth stage comparators (146) receive the difference and identical thresholding constant T of a first order subtracter (145) output respectively separately, if difference is greater than T, output one, otherwise output zero, fourth stage comparator (146) is connected the data that transmit output with first order adder (148); A first order adder (148) receives the data of eight fourth stage comparators (146) output, calculate they and, first order adder (148) is connected with second level subtracter (149) that transmission calculates with; Second level subtracter (149) receive first order adder (148) output and with constant one, should and deduct one, second level subtracter (149) is connected poor that transmission calculates with look-up table (1410); Look-up table (1410) receives the data of second level subtracter (149) output, is that subscript is searched with these data, and look-up table (1410) is connected the value that transmission finds with second level adder (1411); Second level adder (1411) receives the data of look-up table output and the data of second level delayer (147) output, and summation obtains Logsum.
The preparation method of look-up table (1410) is described in the present embodiment code translator LLR computing unit eight input Logsum structures
N wherein
1Value be 0,1..., 7, a is called modifying factor.Described modifying factor a is variable, and value is about 1, and value is 1.1 o'clock in the present embodiment, and performance is best.
Thresholding T is variable described in the present embodiment code translator LLR computing unit eight input Logsum structures, and value is about 4, can not be too big, can not be too little, and in case determine, be exactly a constant, in the present embodiment, it is 4 that T gets a suitable value.Thresholding T chooses, as shown in Figure 6.
The code translator of present embodiment makes forward state metric recursion and back keep minimum complexity to the implementation structure of state measurement recursion, and the calculating implementation structure of LLR adopts eight input Logsum structures of the middle compared to existing technology band correction of eight input Logsum structures provided by the invention, as shown in Figure 3, also reduced complexity.
Present embodiment adopts the BPSK modulation system, encoder is the recursive system convolution coder of eight states, generator polynomial is g0 (D)=1+D2+D3, g1 (D)=1+D+D3, the Turbo code interleaver is the QPP interleaver of the physical layer definition of 3GPP LTE-R8 version, data length is N=1024, and channel adopts awgn channel, sets received signal to noise ratio E
b/ N
0Scope is 0: 0.2: 1.6dB, and just from 0dB, step-length is 0.2dB, stops at the 1.6dB place, four iteration are adopted in decoding.
The simulation result technological progress that present embodiment obtained be significantly, beneficial effect is to give prominence to, from following performance and two aspects of complexity simulation result is described:
The error code result of present embodiment statistics, as shown in Figure 7, wherein " Max-Log-Map " is the forward state metric recursion, the Logsum structure of back in state measurement recursion and LLR calculating all adopts structure as shown in Figure 2; " LLR-Log-Map " is that forward state metric recursion unit and back two input Logsum structures in state measurement recursion unit adopt structure as shown in Figure 2, and the input of eight in LLR computing unit Logsum structure adopts the break-even structure that contains index and logarithm operation; " the present invention " be the structure that adopts of forward state metric recursion unit and back two input Logsum structures in state measurement recursion unit as shown in Figure 2, and the input of eight in LLR computing unit Logsum structure adopts structure provided by the invention; " Log-Map " is forward state metric recursion unit, and the Logsum structure of back in state measurement recursion unit and LLR computing unit all adopts the break-even structure that contains index and logarithm operation; " Cnat-Log-Map " is the forward state metric recursion, and the Logsum structure of back in state measurement recursion and LLR calculating all adopts structure as shown in Figure 3, but the simplest lookup table mode is adopted in tabling look-up here, only contains a constant in the table.
As shown in Figure 7, from the result mistake of the present invention as can be seen again rate compare with break-even " Log-Map " method and differ about 0.1dB, " Cnst-Log-Map " compares and differs about 0.07dB, be better than being better than about " LLR-Log-Map " 0.05dB about " Max-Log-Map " 0.21dB.
Table 1 has provided the computational complexity comparative result of embodiment various implementation structures when code length is 1,024 half iteration.As can be seen from the table, " Log-Map " and " LLR-Log-Map " contains the exponential sum logarithm operation of much more very complexity, actual be to avoid, do not use this structure; The operand of " Cnst-Log-Map " is also very big, contains huge tabling look-up, addition, subtraction number of times; The operand of " Max-Log-Map " is minimum; The operation times of " the present invention " is slightly more than " Max-Log-Map ", but much smaller than " Cnst-Log-Map " and " Log-Map ", the number of comparisons of " Cnst-Log-Map " is " the present invention " 1.5 times, the look-up table number of times is " the present invention " 23 times, the addition number of times is " the present invention " 2.6 times, and the subtraction number of times is " the present invention " 5.5 times.
The complexity of the various implementation structures of table 1Logsum relatively
| The comparison operation number of times | The exponent arithmetic number of times | The logarithm operation number of times | The look-up table number of times | The subtraction number of times | The addition number of times | |
| Max-Log-Map | 23552 | 0 | 0 | 0 | 0 | 0 |
| The present invention | 31744 | 0 | 0 | 1024 | 9216 | 2048 |
| LLR-Log-Map | 23552 | 7168 | 7168 | 0 | 7168 | 14336 |
| Cnst-Log-MAP | 47104 | 0 | 0 | 23552 | 23552 | 11264 |
| Log-map | 23552 | 23552 | 23552 | 0 | 23552 | 47104 |
The performance and the complexity of comprehensive above various Logsum structures, the present invention has obtained and has obtained a fine combination between performance and the complexity, usually maintain an equal level basically on " Max-Log-Map " method complexity that realizes in " the present invention " on the table and the reality, still obtained the gain of 0.2dB.Present embodiment performance and optimum " Log-Map " maintain an equal level, and are better than " Max-Log-Map ", and operation times significantly reduces than " Log-Map " and " Cnst-Log-Map ".
Claims (7)
1. based on the interpretation method of Log-MAP, it is characterized in that during a turbo decodes, may further comprise the steps:
Step 1, receive soft information data, Branch Computed tolerance;
The branch metric that step 2, usefulness calculate comes recursion to calculate forward state metric, needs to realize two input Logsum computings in recursive process, adopts the Max-Log-Map structure herein, and promptly two input Logsum are output as the maximum of two inputs;
Step 3, the forward state metric that recursion is calculated store with RAM, export after treating to be used for LLR calculating after the state measurement recursion calculates again;
The branch metric that step 4, usefulness calculate comes recursion to calculate the back to state measurement, also needs to realize two input Logsum computings in recursive process, still adopts the Max-Log-Map structure herein;
Step 5, the branch metric that will calculate are stored in the forward state metric among the RAM, and recursion calculates afterwards obtains LLR to state metric calculation, will realize eight input Logsum computings in calculating the LLR process;
Step 6, LLR is deducted system information and prior information in the input, calculate external information;
Step 7, the external information back that interweaves is inputed to second sub-decoder as prior information, repeating step 1 restarts a decode procedure to step 6, finishes iteration one time;
Step 8, after the needs number of iterations is finished, LLR is carried out hard decision has just obtained decode results.
2. based on the interpretation method of Log-MAP, it is characterized in that in the turbo decoding as claimed in claim 1 that realization eight input Logsum computings described in the step 5 may further comprise the steps:
The first step is by relatively obtaining eight maximums in the input;
Second step, this maximum subtracted each other with these eight inputs respectively and judge each difference whether in the thresholding of setting, if then export one, otherwise output zero, thresholding herein is a predefined constant;
In the 3rd step, will deduct one again after these output additions;
In the 4th step, the value that obtains is removed to search a good as calculated look-up table as subscript;
The 5th step was exactly the approximation of eight input Logsum with the maximum addition result of searching in eight inputs that the value that obtains and the first step before obtain, and calculated LLR by the values of the eight input Logsum that obtain.
In the turbo as claimed in claim 1 decoding based on the code translator of the interpretation method of Log-MAP, it is characterized in that, comprise: branch metric calculation unit, forward state metric recursion unit, ram cell, back are to state measurement recursion unit and LLR computing unit, the branch metric calculation unit receives the input data, is connected with LLR computing unit three to transmit to state measurement recursion unit with forward state metric recursion unit, back respectively simultaneously to calculate branch metric; Forward state metric recursion unit receives the data of branch metric calculation unit output, forward state metric recursion unit is connected the data that transmit output with ram cell, and temporarily store by ram cell, ram cell is connected with the LLR computing unit that transmit will the forward state metric data of back after the state measurement recursion is calculated; The back is connected with the LLR computing unit to state measurement recursion unit, and the back goes out the back to state measurement to the data recursion that state measurement recursion unit receives the output of branch metric calculation unit; The LLR computing unit receives the branch metric calculation unit, and ram cell and back calculate LLR to the data of state measurement recursion unit output.
4. in the turbo as claimed in claim 3 decoding based on the code translator of Log-MAP, it is characterized in that, the structure of eight input Logsum comprises level Four comparator, two-stage subtracter, two stage adder, two-stage delayer and a look-up table in the described LLR computing unit, eight input data are connected with eight first order delayers with four first order comparators simultaneously, four first order comparators receive eight input data, each comparator receives two input data, and first order comparator is connected the maximum value data that transmits output with second level comparator; Two second level comparators receive four data of first order comparator output, and second level comparator is connected the maximum value data that transmits output with third level comparator; A third level comparator receives two data of second level comparator output, and third level comparator shows eight first order subtracters simultaneously respectively and is connected the maximum value data that transmits output with a second level delayer; A second level delayer receives the data of third level comparator output, with five clocks of this data delay, and second level delayer and data after second level adder is connected propagation delay; Eight first order delayers receive eight input data, and each delayer receives input data and postpones three clocks, and each first order delayer is connected the data that transmit output respectively with a first order subtracter; Eight first order subtracters receive the data of first order delayer and the output of third level comparator, each first order subtracter all receives the data of third level comparator output, receive the data of a first order delayer output simultaneously, first order subtracter is connected poor that transmission calculates with fourth stage comparator; Eight fourth stage comparators receive the difference and identical thresholding constant T of a first order subtracter output respectively separately, if difference greater than T, exports one, otherwise output zero, fourth stage comparator is connected the data that transmit output with first order adder; First order adder receives the data of eight fourth stage comparators outputs, calculate they and, first order adder is connected with second level subtracter that transmission calculates with; Second level subtracter receive the output of first order adder and with constant one, should and deduct one, second level subtracter is connected poor that transmission calculates with look-up table; Look-up table receives the data of second level subtracter output, is that subscript is searched with these data, and look-up table is connected the value that transmission finds with second level adder; Second level adder receives the data of look-up table output and the data of second level delayer output, and summation obtains Logsum.
5. based on the code translator of Log-MAP, it is characterized in that in the turbo decoding as claimed in claim 4 that the preparation method of look-up table is in the described eight input Logsum structures
n
1Value be 0,1..., 7, a is called modifying factor.
6. in the turbo as claimed in claim 5 decoding based on the code translator of Log-MAP, it is characterized in that described modifying factor parameter a refers to when determining described look-up table value, introduces a modifying factor, so that the value of described look-up table is optimum, the value of a is 1.1.
7. based on the code translator of Log-MAP, it is characterized in that in the turbo decoding as claimed in claim 4 that the value of threshold parameter T described in the described eight input Logsum structures is 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105168759A CN101964665B (en) | 2010-10-22 | 2010-10-22 | Log-MAP based decoding method and decoding device thereof in turbo decoding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105168759A CN101964665B (en) | 2010-10-22 | 2010-10-22 | Log-MAP based decoding method and decoding device thereof in turbo decoding |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101964665A true CN101964665A (en) | 2011-02-02 |
| CN101964665B CN101964665B (en) | 2012-11-28 |
Family
ID=43517425
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010105168759A Expired - Fee Related CN101964665B (en) | 2010-10-22 | 2010-10-22 | Log-MAP based decoding method and decoding device thereof in turbo decoding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101964665B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102158235A (en) * | 2011-04-26 | 2011-08-17 | 中兴通讯股份有限公司 | Turbo decoding method and device |
| CN103475380A (en) * | 2013-09-06 | 2013-12-25 | 清华大学 | Parallel Turbo decoding method for image processor |
| CN105634508A (en) * | 2015-12-21 | 2016-06-01 | 西安空间无线电技术研究所 | Realization method of low complexity performance limit approximate Turbo decoder |
| WO2023116504A1 (en) * | 2021-12-20 | 2023-06-29 | 大连海事大学 | Double-factor correction turbo decoding method based on simulated annealing algorithm |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1744444A (en) * | 2005-09-30 | 2006-03-08 | 上海交通大学 | Dual binary turbo code decoding method based on adaptive principle |
| CN1983826A (en) * | 2005-12-15 | 2007-06-20 | 大唐移动通信设备有限公司 | Turbo decoder and decoding method |
-
2010
- 2010-10-22 CN CN2010105168759A patent/CN101964665B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1744444A (en) * | 2005-09-30 | 2006-03-08 | 上海交通大学 | Dual binary turbo code decoding method based on adaptive principle |
| CN1983826A (en) * | 2005-12-15 | 2007-06-20 | 大唐移动通信设备有限公司 | Turbo decoder and decoding method |
Non-Patent Citations (1)
| Title |
|---|
| 《上海交通大学学报》 19980630 徐友云 等 Turbo/MAP编译码器中的交织器优化设计 第43-47页 1-7 第32卷, 第6期 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102158235A (en) * | 2011-04-26 | 2011-08-17 | 中兴通讯股份有限公司 | Turbo decoding method and device |
| WO2012146124A1 (en) * | 2011-04-26 | 2012-11-01 | 中兴通讯股份有限公司 | Turbo decoding method and device |
| RU2571597C2 (en) * | 2011-04-26 | 2015-12-20 | ЗетТиИ Корпорейшн | Turbocode decoding method and device |
| CN102158235B (en) * | 2011-04-26 | 2016-11-23 | 中兴通讯股份有限公司 | The method and device of turbo decoding |
| CN103475380A (en) * | 2013-09-06 | 2013-12-25 | 清华大学 | Parallel Turbo decoding method for image processor |
| CN105634508A (en) * | 2015-12-21 | 2016-06-01 | 西安空间无线电技术研究所 | Realization method of low complexity performance limit approximate Turbo decoder |
| CN105634508B (en) * | 2015-12-21 | 2019-03-26 | 西安空间无线电技术研究所 | A kind of implementation method of the Turbo decoder of the nearly performance limit of low complex degree |
| WO2023116504A1 (en) * | 2021-12-20 | 2023-06-29 | 大连海事大学 | Double-factor correction turbo decoding method based on simulated annealing algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101964665B (en) | 2012-11-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| AU2003247190B2 (en) | Apparatus and method for reducing Bit error rates (BER) and frame error rates (FER) using turbo decoding in a digital communication system | |
| CN108847848B (en) | BP decoding algorithm of polarization code based on information post-processing | |
| CN105634508A (en) | Realization method of low complexity performance limit approximate Turbo decoder | |
| CN101964665B (en) | Log-MAP based decoding method and decoding device thereof in turbo decoding | |
| Ma et al. | A joint early detection-early stopping scheme for short-frame turbo decoding | |
| CN101425870B (en) | Iterative decoding method for Turbo code | |
| KR101923701B1 (en) | Apparatus and method for iterative detection and decoding in wireless communication system | |
| Reddy et al. | A low complexity stopping criterion for reducing power consumption in turbo decoders | |
| Sujatha et al. | Performance improvement of Turbo decoder using VLSI optimization Techniques | |
| CN103701475B (en) | Decoding method for Turbo codes with word length of eight bits in mobile communication system | |
| Zhanji et al. | A new parity-check stopping criterion for turbo decoding | |
| CN107968697B (en) | Decoding method and device for overlapping multiplexing system | |
| CN113258937B (en) | Component decoder, extrinsic information storage unit, and Turbo code decoder | |
| Song et al. | The implementation of turbo decoder on DSP in W-CDMA system | |
| CN113824452B (en) | Decoding method based on grid graph, component decoder and channel decoder | |
| Sharma et al. | Performance analysis of log-map, SOVA and modified SOVA algorithm for turbo decoder | |
| Zhang et al. | Integrated implementation of fixed-point turbo decoders for lte system | |
| Wang et al. | Research on Parallel Turbo Encoding and Decoding Technology | |
| JP5567216B2 (en) | Method and iterative turbo decoder for stopping iteration in an iterative turbo decoder | |
| CN115514377A (en) | Turbo decoding circuit and decoding method | |
| CN1135005C (en) | Outer assigned information treating method of decoding cascade convolution code | |
| Kene et al. | Iterative Decoding Termination Schemes for Turbo Code Performance Optimization In Mobile Wi-Max Environment | |
| Wang et al. | A Novel Decoding Algorithm for BICM-ID Embedded Turbo Codes | |
| CN116232342A (en) | Turbo code decoding method | |
| Lin et al. | A low-complexity stopping criterion for turbo decoding using forward state metrics at a single time instant |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121128 Termination date: 20151022 |
|
| EXPY | Termination of patent right or utility model |