CN1135005C - Outer assigned information treating method of decoding cascade convolution code - Google Patents
Outer assigned information treating method of decoding cascade convolution code Download PDFInfo
- Publication number
- CN1135005C CN1135005C CNB001214144A CN00121414A CN1135005C CN 1135005 C CN1135005 C CN 1135005C CN B001214144 A CNB001214144 A CN B001214144A CN 00121414 A CN00121414 A CN 00121414A CN 1135005 C CN1135005 C CN 1135005C
- Authority
- CN
- China
- Prior art keywords
- decoder
- information
- decoding
- convolution code
- cascade convolution
- 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.)
- Expired - Lifetime
Links
Images
Landscapes
- Error Detection And Correction (AREA)
- Detection And Correction Of Errors (AREA)
Abstract
The present invention relates to an outer assigned information treating method of decoding a cascade convolution code, which carries out a certain compensation for performance loss caused by the simplification of the algorithm of decoders. In a feedback structure of a cascade convolution code decoder, feedback information passed from one decoder to the other decoder is treated by an optimal outer assigned information factor, the feedback information output by the decoders is multiplied by the optimal constant factor, and shift and addition calculation is performed. The optimal constant factor is obtained in constant factor search by a search method in view of the algorithm of the cascade convolution code decoder, and the variation range of the optimal constant factor is from 0.4 to 1.0. The outer assigned information treating method of decoding a cascade convolution code has small performance loss when decoded speed is greatly enhanced, and the outer assigned information treating method of decoding a cascade convolution code also simplifies realization complexity in a certain program.
Description
Technical field
The present invention relates to a kind of data, mobile communication technology, relate to a kind of forward direction (FEC) channel error correction encoding class technology cascade convolution code (Turbo-code) class feedback decoding method or rather, is a kind of processing method to extrinsic information in the decoding.
Background technology
Cascade convolution code (Turbo-code), it is a kind of very novel channel coding and decoding method that proposes in 1993 by the Frenchman earlier, under certain condition, only differ about 0.7dB with Shannon (Shannon) limit, because of himself very excellent performance, therefore once appearance, just cause the sensation of whole coding circle, some scholar even to estimate it be revolution in the channel coding technology field.Given this, International Telecommunications Union (ITU-International Telecommunication Union) also classifies it as the standard of 3-G (Generation Three mobile communication system) (IMT-2000) coding field.The basic principle of Turbo-code coding and decoding can be referring to Fig. 1, Fig. 2.
Referring to Fig. 1, has the Turbo-code coder structure of 1/2,1/3 code efficiency shown in the figure.Input data information also is system information X.Input data information X forms one-component sign indicating number information Y (redundant digit) through the effect of systematic code convolution coder-rsc encoder 1, and input data information X forms another component code information Y ' (redundant digit) through the effect of interleaver and systematic code convolution coder-rsc encoder 2.Follow-up perforation unit only deals with output information Y, the Y ' of two systematic code convolution coder-rsc encoders 1, rsc encoder 2, and information after the processing and system information X be outputting encoded data after the multiplexer effect again, obtains desired code efficiency.
Referring to Fig. 2, typical Turbo-code double-core iterative decoder structure shown in the figure, receive data RK and revert to three partial information RX, RY, RY ' through separating perforation unit, wherein two respectively output information Y, the Y ' of corresponding two rsc encoders, and send two soft inputs/soft output (SISO) decoder-SISO decoder 1 and SISO decoder 2 (forming the double-core iteration structure) wherein to also have the process of interleaving/de-interleaving (finishing) respectively by interleaver and deinterleaver.
The decode procedure of finishing repeatedly iteration of cooperating mutually between this typical Turbo-code double-core iterative decoder is: beginning is deciphered computing by SISO decoder 1 earlier, its input data are RX, RY and the extrinsic information Z2 that exports by SISO decoder 2, (this extrinsic information Z2 perseverance is 0 when Initial Decoding), the decoding computing of SISO decoder 1 finishes new extrinsic information Z1 of back output, behind the interleaving effect of interleaver, form information Z1 ', and send into SISO decoder 2 as prior information, that sends into SISO decoder 2 simultaneously also has through separating information RY ' that perforation unit recovers and recovering (RX) and the information RX ' behind the interleaver interleaving effect through separating perforation unit, after the decoding computing of SISO decoder 2 finishes, can obtain a new extrinsic information, after the effect of deinterleaving of deinterleaver, this extrinsic information Z2 is admitted to SISO decoder 1, finish once complete iterative process, repeat to begin said process then, so move in circles, till when satisfying systematic function, therefore whole decode procedure is exactly a repeatedly process of iteration.Another likelihood value information that SISO decoder 2 produces among Fig. 2 is used for judgement after the effect of deinterleaving of deinterleaver, produce the output after error correction.
In the above-mentioned decode procedure, because of extrinsic information Z1, Z2 are fed use, promptly be called feedback decoding, the benefit of feedback decoding is fully to excavate the contribution of redundant information to system information, but then, in order to prevent to introduce positive feedback, usually need extrinsic information Z1, the Z2 of feedback are done certain processing.
On the other hand, the difference of the decoding algorithm that is adopted according to the SISO decoder, the Turbo-code decoder has three kinds in MAP class decoder, MAX-LOG-MAP class decoder and SOVA class decoder again.No matter be which kind of decoder, because the decoding algorithm (the standard decoding algorithm refers to the MAP algorithm) of Turbo-code is very complicated, required operand is very big, the decoding algorithm that will realize the Turbo-code under the higher rate in real time almost is an impossibility, therefore when specific implementation, all to carry out to a certain degree simplification to algorithmic formula, could with circuit realize one near available Turbo-code compile, decoder, bring certain performance loss therefrom.If with standard MAP performance is benchmark, the performance loss of MAX-LOG-MAP class decoder and SOVA class decoder is respectively about 0.5dB, 1.0dB, and in the occasion of some performance requirement strictnesses, so the performance loss of " greatly " is insufferable.Given this, just must adopt the way of some " compensation " to come " drawing " to return some performance losss, its main method is done " fitting " to channel exactly, promptly between per twice iteration (decoding), carry out extrinsic information and handle (also can claim compression), thereby the part compensation is because of simplifying the performance loss that decoding algorithm causes.
Extrinsic information commonly used is handled (compression) method two kinds: Gauss's approximatioss and log-compressed method.But Gauss's approximatioss list of references [1] Berrou wherein, Near Optimum Error CorrectingCoding And Decoding:Turbo-Codes, IEEE Transactions on Communications, Vol.44, NO.10, October 1996..; But log-compressed method list of references [2] Berrou wherein, Error-Correction coding method with at least two systematicconvolutional codings in parallel, corresponding iterative decodingmethod, decoding module and decoder.United States Patent, patent number:5.446,747, Aug.29,199 5., and document [3] Berrou et al.Method for a maximumlikelihood decoding of a convolutional code with decision weighting, andcorresponding decoder, United States Patent, patent number:5,406,570, Apr.11,1995..Performance with the log-compressed method among these two kinds of methods is good.
In conjunction with referring to Fig. 3, have the Turbo-code decoder architecture that extrinsic information is handled shown in the figure, and the difference of the maximum of Turbo-code decoder architecture shown in Figure 2 is: between SISO decoder 1 and SISO decoder 2 and SISO decoder 2 and SISO decoder 1, adopt Gauss's approximatioss or log-compressed method that the extrinsic information between twice iteration is handled, simplify the performance loss problem that causes because of algorithm thereby reach the part compensation.
After the extrinsic information processing through Gauss's approximatioss or log-compressed method,, brought problem on the other hand thus though can retrieve the performance loss of about 0.2~0.3dB: complex disposal process; Especially Gauss's approximatioss, owing to also to " know " channel statistic property, and this point is to be difficult to obtain in real system.Remove outside above two reasons, to realize accurately in hardware system that Gauss's approach method and log-compressed method also are extremely difficult, Gauss's approximatioss need be made division arithmetic, the log-compressed method need be made multiplication and division, ask the computing of log, obviously, go to realize that with hardware the extremely long complex calculations of this figure place are quite difficult things.
Adopt Gauss's approximatioss to be in the document [1]: to suppose that extrinsic information is that X (is a soft value to extrinsic information processing method, be non integer value), hypothesis X satisfies Gauss (Gauss) distribution earlier, so-called Gauss approaches and exactly X is multiplied by a variable relevant with the characteristic of channel, thereby obtains revised " newly " channel value: (2m/ σ
2) * X, wherein: m is average, σ
2Be variance.Generally speaking, when being, channel becomes, i.e. (2m/ σ
2) will change along with the variation of time (comprising Frame and iteration), therefore each iteration and every frame all will calculate this value, quite bother, and effect are unsatisfactory.
The log-compressed method that adopts in document [2], [3], be about to the form that extrinsic information X becomes d * log (1+|X|/d), wherein d is constant (extrinsic information that comes down to a frame take absolute value after mean value), adopt the performance of this method a kind of more better than preceding, but calculating more complicated, average and do log computing etc. as needs, though in the realization of side circuit, can consider to adopt the mode of tabling look-up, but still need to average, can bring the expense of extra memory simultaneously.
In a word, no matter adopt any in above two kinds of methods, calculate cumbersome as a whole, and along with the increase of signal to noise ratio, " floor (Floor) effect " also can more early appear, be that signal to noise ratio is increased to a certain degree, even increase considerably signal to noise ratio again, the error rate of system can descend very slow even the rising phenomenon occur.Specifically, so-called floor effect, be meant in Turbo-code decoding, immobilize that signal to noise ratio (snr) is from little change when big in other condition, the decoding performance of Turbo-code can be become better and better, be that the error rate (BER) is more and more littler, but when signal to noise ratio (snr) reached certain value, the improvement of the error rate (BER) was just not obvious, a tangible gentle slope can occur, this phenomenon just is called as floor (Floor) effect.
Summary of the invention
Shortcoming in view of above existing outer assigned information treating method, the objective of the invention is to propose the outer assigned information treating method that a kind of new cascade convolution code (Turbo-code) is decoded, make realize simple and performance better, utilize this extrinsic information treating method can simpler, more practicably realize the fast decoding of Turbo-code, the while can reach the performance loss that brings because of shortcut calculation and be better than the existing methods compensation.
The object of the present invention is achieved like this: a kind of outer assigned information treating method of decoding cascade convolution code, the extrinsic information 1 of decoder 1 output feeds back to decoder 2 and handles after first extrinsic information is handled, the extrinsic information 2 of decoder 2 outputs feeds back to decoding 1 and handles after second extrinsic information is handled, form repeatedly the decode procedure of iteration, it is characterized in that: it is the feedback information that multiply by decoder 1 output with best invariant that described first extrinsic information is handled, and it is the feedback information that multiply by decoder 2 outputs with best invariant that described second extrinsic information is handled.
Described invariant with the best is multiplied by feedback information and does shift operation and add operation exactly.
The invariant of described the best, be at the decoding cascade convolution code device algorithm, carry out invariant search by search method and obtain; Described search method is the variation along with invariant, makes ber curve present approximate " V " font and changes, and the pairing invariant of " V " font bottom minimum bit-error rate of ber curve is exactly an invariant best under the minimum bit-error rate.
Described decoding cascade convolution code device comprises Parallel Concatenated Convolutional Code decoder and Serial Concatenated Convolutional Code decoder, the excursion of described best constant factor is taken as 0.4-1.0, and this best constant factor is that shortcut calculation with described decoding cascade convolution code device changes.
The outer assigned information treating method of the decoding cascade convolution code that the present invention proposes, be actually a kind of and after simplifying decoder algorithms, its performance done the to a certain degree method of compensation, this method is simple, be that a kind of available hardware circuit is realized, can practical Turbo-code decoder.After the Turbo-code decoder adopts the algorithm of simplifying and uses outer assigned information treating method of the present invention, when improving decoding speed greatly, and performance loss is also little, as with respect to standard MAP algorithm, performance is only lost about 0.15dB (in code length=2000, code efficiency=1/3, iteration 6 times is under the condition that on average interweaves).Therefore, from complexity and performance loss compromise, the present invention is a method with beneficial effect, and its final disposal ability will be much higher than 2.048Mbps (also being the highest access rate of user in the 3GPP-Third Generation Partnership Project standard).
Adopt best constant factor method of the present invention that extrinsic information is multiplied each other after (compression), can will reduce to bottom line to extrinsic information processing complexity, the software and hardware circuit of being convenient in the real system is realized; Can also improve simultaneously the decoding performance of Turbo-code, as the gain about 0.1dB can be arranged with respect to the processing method in the background document, for the decoder that does not adopt any processing method, the gain of 0.3-0.4dB is nearly arranged, also can reduce " floor " (Floor) influence of effect; Further analysis also shows, adopt method of the present invention after, also help the convergence and the robustness that improve iteration, make this Turbo-code coding more can adapt to complicated wireless channel.
Description of drawings
Fig. 1 is the structural representation of Turbo-code encoder in the background technology.
Fig. 2 is the structural representation of Turbo-code decoder in the background technology.
Fig. 3 is the structural representation that has the Turbo-code decoder of extrinsic information processing in the background technology.
Fig. 4 is the structural representation that the present invention has the Turbo-code decoder of extrinsic information processing unit.
Fig. 5 will adopt the application structure schematic diagram of Turbo-code decoder application in 3-G (Generation Three mobile communication system) reception link of the inventive method.
Fig. 6 be 3066 at code length, signal to noise ratio (snr) is 1.4dB, under 1/2 code efficiency, different invariant, adopt the error rate (BER) curve of the parallel Turbo-code decoder of soft output Viterbi (SOVA) algorithm.
Fig. 7 be 3066 at code length, signal to noise ratio (snr) is 1.4dB, under 1/2 code efficiency, different invariant, adopt the error rate (BER) curve of the parallel Turbo-code decoder of BCJR-MAP algorithm.
Fig. 8 be 3066 at code length, signal to noise ratio (snr) is 1.0dB, under 1/2 code efficiency, different invariant, adopt the error rate (BER) curve of the parallel Turbo-code decoder of BCJR-MAP algorithm.
Fig. 9 be 3066 at code length, signal to noise ratio (snr) is 1.2dB, under 1/2 code efficiency, different invariant, adopt the error rate (BER) curve of the parallel Turbo-code decoder of BCJR-MAP algorithm.
Figure 10 be 3066 at code length, signal to noise ratio (snr) is 1.3dB, under 1/2 code efficiency, different invariant, adopt the error rate (BER) curve of the parallel Turbo-code decoder of maximum MAP (MAX-LOG-MAP) algorithm of taking the logarithm.
Figure 11 be 1000 at code length, signal to noise ratio (snr) is 0.6dB, under 1/3 code efficiency, different invariant, adopt the error rate (BER) curve of the parallel Turbo-code decoder of maximum MAP (MAX-LOG-MAP) algorithm of taking the logarithm.
Figure 12 be 1000 at code length, signal to noise ratio (snr) is 0.6dB, under 1/3 code efficiency, different invariant, adopt the error rate (BER) curve of the serial Turbo-code decoder of BCJR-MAP algorithm.
Figure 13 be 1000 at code length, signal to noise ratio (snr) is 1.0dB, under 1/3 code efficiency, different invariant, adopt the error rate (BER) curve of the serial Turbo-code decoder of MAX-LOG-MAP algorithm.
Figure 14 is is 3066,1/3 code efficiency at code length, adopt the MAX-LOG-MAP algorithm, under the different signal to noise ratio (snr) condition, the error rate (BER) the performance comparison curves of the Turbo-code decoder when adopting the log-compressed method of best constant factor method of the present invention and employing background technology.
Figure 15 is is 3066,1/2 code efficiency at code length, adopt the BCJR-MAP algorithm, under the different signal to noise ratio (snr) condition, the error rate (BER) the performance comparison curves of the Turbo-code decoder when adopting the log-compressed method of best constant factor method of the present invention and employing background technology.
Figure 16 be 2000 at frame length, under the iteration 6 times, different signal to noise ratio condition, the error rate (BER) the performance comparison curves of the Turbo-code decoder when adopting best constant factor method of the present invention and adopting the log-compressed method of background technology.
Embodiment
Further specify technology of the present invention below in conjunction with embodiment and accompanying drawing.
Referring to Fig. 4, adopt the new outer assigned information treating method of the present invention Turbo-cod decoder model afterwards shown in the figure.Core cell among Fig. 4 has: algorithm nuclear 2 (SISO decoder DEC2), the interleaver of the algorithm of SISO nuclear 1 (SISO decoder DEC1), SISO, remove (separating) interleaver, two extrinsic information processing units and decision unit.SISO decoder DEC1 and SISO decoder DEC2 can adopt standard MAP, suboptimum MAP (MAX-LOG-MAP), SOVA scheduling algorithm, but are not limited to these algorithms.The output 11 of the algorithm of SISO nuclear 1 (SISO decoder DEC1) through the effect of an extrinsic information processing unit after output information 12, information 12 is examined the prior information input of 2 (SISO decoder DEC2) again as the algorithm of another SISO behind the interleaver interleaving effect, the output of the algorithm nuclear 2 (SISO decoder DEC2) of SISO is through the deinterleaver effect with through the output information after another extrinsic information processing unit effect, import SISO algorithm nuclear 1 (SISO decoder DEC1) again, so iterate down (can be as required during enforcement and desired hardware size, adopt a plurality of such SISO algorithms nuclears to feed back connection).
Core content of the present invention is exactly to adopt best constant factor to multiply by the extrinsic information of decoder output, substitutes complicated algorithm commonly used the extrinsic information of decoder output is handled.For the different factors that influence decoding performance, for example: coder structure, encoding constraint length, frame length, different decoder algorithms etc., the range of choice of invariant can be different, but analysis and territory are tested and shown: these are through simplifying the invariant that all there is an optimum in the algorithms of realizing, under this best constant factor, bit error rate performance best (other condition identical situation under).Can obtain the best constant factor under the different situations by the method for computer search, accompanying drawing 6 to accompanying drawing 13 illustrates this computer search process.
The Turbo-code decoder is shown respectively among Fig. 6 to Figure 13 adopts SOVA, BCJR-MAP, MAX-LOG-MAP, parallel/serial Turbo-code scheduling algorithm, the error rate (BER) performance curve under different invariant effects, represent the variation of invariant with X-axis, be illustrated in the error rate under the various invariants with Y-axis, as can be seen from the figure: these curve charts are " V " font substantially to be changed; All having a performance in these curve charts preferably is the point of BER minimum, i.e. the pairing invariant of point of the bottom BER minimum of " V " font is just corresponding to the best constant factor under this kind condition.No matter under which kind of situation, the scope of this best constant factor is between 0.4~1.0.Best constant factor shown in Fig. 6, Fig. 7,10 is 0.7, is 0.9 as Fig. 8, Fig. 9, best constant factor shown in Figure 12, and best constant factor as shown in figure 13 is about 0.5, and best constant factor as shown in figure 11 is 0.6-0.8." V " font curve shown in Figure 7, its right half part almost becomes flat, invariant is also arranged with the different slightly different situations of signal to noise ratio in other accompanying drawing, but without prejudice in conjunction with the conclusion of carrying out aggregate analysis than the low signal-to-noise ratio situation, and further say, also can be according to this invariant of different adjustment of signal to noise ratio.
By above-mentioned listed Search Results with the acquisition of computer searching method, can prove the best constant factor that certain existence is such, can be so that decoding performance under certain condition reaches best.
Figure 14 and Figure 15 comparative descriptions adopt best constant factor method of the present invention and adopt the log-compressed method of background technology, the performance of Turbo-code decoding is relatively.Represent signal to noise ratio with X-axis, represent the error rate with Y-axis.
What lighter curve was represented among Figure 14 is that the error rate that adopts best constant factor method of the present invention to obtain changes, what darker curve was represented is to adopt the error rate of the log-compressed method acquisition of background technology to change, both are more as can be seen: under same signal to noise ratio condition, adopt the performance of the outer assigned information treating method of invariant to be better than the processing method that adopts log-compressed, and processing procedure is very simple.
What lighter curve among Figure 15 (solid line) was represented is that the error rate that adopts best constant factor method of the present invention to obtain changes, what darker curve (dotted line) was represented is to adopt the error rate of the log-compressed method acquisition of background technology to change, both are more as can be seen: under same signal to noise ratio condition, adopt the performance of the outer assigned information treating method of invariant to be better than the processing method that adopts log-compressed, and processing procedure is very simple.
Further emulation shows: this invariant processing method is effective equally to serial Turbo-code.And adopt the method for this compression extrinsic information, can also reduce " floor (Floor) effect " effectively.
Figure 16 illustrates the log-compressed method that adopts best constant factor method of the present invention and adopt background technology, to the influence of floor (Floor) effect.Represent signal to noise ratio (snr) with X-axis among the figure, represent the error rate (BER) with Y-axis.
Relatively adopt the error rate (BER) curve (representing) of best constant factor method acquisition of the present invention among Figure 16 with light line, the error rate (BER) curve (representing) with the log-compressed method acquisition of adopting background technology with dark line, along with the increase of signal to noise ratio, adopt the time of the floor effect of best constant factor method appearance will be later than the log-compressed method that adopts background technology.
Referring to Fig. 5, the application example of high-speed Turbo-code decoder in the third generation (3G) mobile communication system of realizing with the inventive method shown in the figure.In/receiving element that coupler form multiple by front radio-frequency unit, demodulation/despreader, analog/digital converter, Turbo-code decoder and base band, the Turbo-code decoder just can adopt the method with best constant factor of the present invention to handle extrinsic information, and designs the Turbo-code decoder product of commercialization.
Method of the present invention is: in the decoder of Turbo-code, adopt the method that multiply by extrinsic information with best constant factor to replace Gauss to approach or the method for log-compressed; In the realization circuit of reality, only need simply the extrinsic information of deciphering output to be multiply by this best constant factor and get final product; The range of choice of this best constant factor is 0.4-1.0, changes with different decoding shortcut calculations; The present invention is applicable to MAP, LOG-MAP (the maximum a posteriori algorithm of taking the logarithm), MAX-LOG-MAP, SOVA, serial Turbo-code decoder to extrinsic information processing way.
Claims (10)
1. the outer assigned information treating method of a decoding cascade convolution code, the extrinsic information 1 of decoder 1 output feeds back to decoder 2 and handles after first extrinsic information is handled, the extrinsic information 2 of decoder 2 outputs feeds back to decoding 1 and handles after second extrinsic information is handled, form repeatedly the decode procedure of iteration, it is characterized in that: it is the feedback information that multiply by decoder 1 output with best invariant that described first extrinsic information is handled, and it is the feedback information that multiply by decoder 2 outputs with best invariant that described second extrinsic information is handled.
2. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 1 is characterized in that: describedly be multiplied by feedback information with best invariant and do shift operation and add operation exactly.
3. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 1 is characterized in that: the invariant of described the best, be at the decoding cascade convolution code device algorithm, carry out the invariant search by search method and obtain; Described search method is the variation along with invariant, makes ber curve present approximate " V " font and changes, and the pairing invariant of " V " font bottom minimum bit-error rate of ber curve is exactly an invariant best under the minimum bit-error rate.
4. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 3, it is characterized in that: described decoding cascade convolution code device comprises Parallel Concatenated Convolutional Code decoder and Serial Concatenated Convolutional Code decoder, the excursion of described best constant factor is taken as 0.4-1.0, and this best constant factor is that shortcut calculation with described decoding cascade convolution code device changes.
5. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 4 is characterized in that: when the shortcut calculation of described Parallel Concatenated Convolutional Code decoder was soft output Viterbi algorithm, described best constant factor was 0.7.
6. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 4, it is characterized in that: when the shortcut calculation of described Parallel Concatenated Convolutional Code decoder was the BCJR-MAP algorithm, described best constant factor was 0.7-0.9.
7. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 4, it is characterized in that: when the shortcut calculation of described Parallel Concatenated Convolutional Code decoder was the maximum MAP algorithm of taking the logarithm, described best constant factor was 0.6-0.8.
8. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 4, it is characterized in that: when the shortcut calculation of described Serial Concatenated Convolutional Code decoder was the BCJR-MAP algorithm, described best constant factor was 0.9.
9. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 4, it is characterized in that: when the shortcut calculation of described Serial Concatenated Convolutional Code decoder was the maximum MAP algorithm of taking the logarithm, described best constant factor was 0.5.
10. the outer assigned information treating method of a kind of decoding cascade convolution code according to claim 1 is characterized in that: the described information that feeds back to decoder 2 is through the information after the interleaving treatment; The described information that feeds back to decoder 1 is the information through deinterleaving after handling.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB001214144A CN1135005C (en) | 2000-07-20 | 2000-07-20 | Outer assigned information treating method of decoding cascade convolution code |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB001214144A CN1135005C (en) | 2000-07-20 | 2000-07-20 | Outer assigned information treating method of decoding cascade convolution code |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1335731A CN1335731A (en) | 2002-02-13 |
CN1135005C true CN1135005C (en) | 2004-01-14 |
Family
ID=4588786
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB001214144A Expired - Lifetime CN1135005C (en) | 2000-07-20 | 2000-07-20 | Outer assigned information treating method of decoding cascade convolution code |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1135005C (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105531938A (en) * | 2013-09-10 | 2016-04-27 | 华为技术有限公司 | A turbo decoder for decoding an input signal |
-
2000
- 2000-07-20 CN CNB001214144A patent/CN1135005C/en not_active Expired - Lifetime
Also Published As
Publication number | Publication date |
---|---|
CN1335731A (en) | 2002-02-13 |
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 | |
US6298461B1 (en) | Encoding and decoding methods and apparatus | |
CN1203616C (en) | TURBO decoder with decision feedback equalization | |
CN1178399C (en) | Highly parallel MAP decoder | |
JP4709119B2 (en) | Decoding device and decoding method | |
EP1678834A1 (en) | Turbo code decoder with parity information update | |
JP2003528477A (en) | Partitioned deinterleaver memory for map decoder | |
CA2366592A1 (en) | A system and method employing a modular decoder for decoding turbo and turbo-like codes in a communications network | |
CN1188951C (en) | Scaled-feedback turbo decoder | |
CN105141394A (en) | Efficient soft value generation for coded bits in a turbo decoder | |
Papaharalabos et al. | SISO algorithms based on Max-Log-MAP and Log-MAP turbo decoding | |
CN1135005C (en) | Outer assigned information treating method of decoding cascade convolution code | |
Halter et al. | Reconfigurable signal processor for channel coding and decoding in low SNR wireless communications | |
US7634703B2 (en) | Linear approximation of the max* operation for log-map decoding | |
EP1094612B1 (en) | SOVA Turbo decoder with decreased normalisation complexity | |
CN1145266C (en) | Turbo code decoding method and decoder | |
CN1204693C (en) | Stochastic system Turbo code coding and decoding method | |
CN1338824A (en) | Method and decoder for decoding turbo code | |
Parvathy et al. | Throughput enhancement of SISO parallel LTE turbo decoders using floating point turbo decoding algorithm | |
CN2884696Y (en) | Normalized device for Turbo coder decoder | |
CN100505600C (en) | An implementing method for shortening critical path of Turbo decoder | |
CN100571097C (en) | A kind of method that is used for Turbo code decoder minimizing storage resources | |
Yang et al. | Implementation of Turbo code based on CCSDS | |
Wan et al. | A novel parallel turbo coding technique based on frame split and trellis terminating | |
Wang et al. | Research on Parallel Turbo Encoding and Decoding Technology |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C06 | Publication | ||
PB01 | Publication | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CX01 | Expiry of patent term | ||
CX01 | Expiry of patent term |
Granted publication date: 20040114 |