CN1968023A - A coding method of convolution code - Google Patents
A coding method of convolution code Download PDFInfo
- Publication number
- CN1968023A CN1968023A CN 200610087453 CN200610087453A CN1968023A CN 1968023 A CN1968023 A CN 1968023A CN 200610087453 CN200610087453 CN 200610087453 CN 200610087453 A CN200610087453 A CN 200610087453A CN 1968023 A CN1968023 A CN 1968023A
- Authority
- CN
- China
- Prior art keywords
- path
- sliding window
- moment
- preferred
- recalling
- 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 convolutional code decode method, wherein it leads in sliding window technique into traditional PLVA; segments convolutional codes into different sliding windows; searches local optimized path in each sliding window; based on the local optimized path, obtains globe optimized path; checks all globe optimized paths; and outputs decoded result. The invention can reduce occupied space of PLVA decode.
Description
Technical field
The present invention relates to the interpretation method field, particularly relate to a kind of coding method of convolution code.
Background technology
Convolution code is a kind of channel coding method that Ai Lisi (Elias) proposed in nineteen fifty-five.The verification unit of convolution code is information-related with this group not only, and relevant with the information sets that inputed to encoder in the past.Equally, also will information extraction from the relevant code character in front and back in the decode procedure, therefore, convolution code has reasonable letter energy, and because coding is simple, coding gain is high and have very strong correction random error ability, therefore obtains very using widely in communication system.
Viterbi algorithm (Viterbi Algorithms, be called for short VA) be a kind of based on the maximum-likelihood criterion algorithm on the grid chart basis of sign indicating number, it also is the coding method of convolution code of performance the best under additive white Gaussian noise channel, owing to have efficient height, fast, the decoding characteristic of simple of speed, therefore be widely used in the various communication systems simultaneously.
Based on VA, Texas Instruments company has announced the document (hereinafter to be referred as document 1) of one piece of exercise question for " viterbi decoder coprocessor TMS320C64x DSP reference guide " (TMS320C64xDSPViterbi-Decoder Coprocessor (VCP) Reference Guide ") in September, 2004.The solution of the interpretation method that a kind of VA of application deciphers has been proposed in the literature.
Its technical scheme specifically is to have introduced the technology of sliding window in traditional VA, to reduce the required memory cell of decoding.According to the characteristic of VA, when decoding depth d is enough big (such as, d=4～5*K, wherein K is the constraint length of convolution coder), the probability that all survivor path meetings are very big merges in d chronomere.When frame length is enough big, the segmentation of convolution frame in sliding window.If the width of sliding window is W=R+C, wherein the R section is the judgement output, and the C section is the lap of adjacent two sliding windows.State measurement and recall in whole slide window and carry out is only exported R judgement.By this mechanism, when frame length was longer, the memory cell that can reduce the path register significantly took.Specifically being, is M for a convolutional encoding block length, and status number is the convolution code of N, uses traditional VA, and the required memory cell that takies is: M*N state storage unit, and M*N tolerance memory cell; And use document 1 described technical scheme, needed memory cell is: W*N state storage unit, and W*N tolerance memory cell.As seen, the technical scheme described in the document 1, the W/M that the required memory space that takies is traditional VA is doubly.
Simultaneously, expansion based on VA, Nambirajian Seshadri and Carl-Erik W.Sundberg are at U.S.'s electronic motor Ssociety of engineers journal of communicating by letter, 1994, February, March, April, the 42nd volume, the exercise question of 313-323 page or leaf is " tabulation Viterbi decoding algorithm and application " document (Nambirajan Seshadri andCarl-Erik W.Sundberg, List Viterbi Decoding Algorithms with Applications, IEEETrans.on Comm., Vol.42, No.2/3/4, Feb./March/April 1994, proposed a kind of tabulation viterbi algorithm (generally being also referred to as sub-optimal path algorithm (List VA is called for short LVA)) in 313-323.).Fig. 1 is the decoding architecture schematic diagram of this method, and as shown, LVA from L globally-optimal selection path, selects most possible correct path according to the result of error checking and correction in decode procedure.Because the relative VA of LVA has increased (L-1) individual preferred path in the search procedure, has improved the probability that finds correct decoding, is widely used in comprising in the decoding system of convolution code and check code.
Parallel sub-optimal path algorithm (Parallel LVA is called for short PLVA) is a kind of basic implementation of LVA.Figure 2 shows that the method schematic diagram of PLVA, as shown, the basic implementation of PLVA is: according to VA, by a grid search, obtain L globally-optimal selection path simultaneously.Owing to all find L the local preferred path that enters this state for each each state constantly in the grid chart, PLVA is at each each state constantly, calculate 2 branch metrics and 2L path metric, finally once obtain needed L globally-optimal selection path.Special when L=1, be exactly VA.
But, PLVA all calculates 2 branch metrics and 2L path metric at each each state constantly, simultaneously, need the matrix of L M*N to store L preferred path of each every state point correspondence constantly respectively, wherein M is the convolutional encoding block length, N is the state number, and L is the total number of paths in the default globally-optimal selection path that obtains.Compare with VA, this method needs the individual path metric of extra computation 2 (L-1), additionally takies (L-1) individual M*N matrix stores preferred path.The amount of calculation of PLVA and memory space approximately are L times of VA algorithm, and basic and L is the relation of growth.Therefore, though PLVA has improved the probability that finds correct decoding,, limited the practical application of PLVA greatly because it need take huge memory space.
Summary of the invention
The technical problem to be solved in the present invention provides a kind of coding method of convolution code, and PLVA need take the problem of excessive memory space in the existing method of solution.
For solving the problems of the technologies described above, the objective of the invention is to be achieved through the following technical solutions:
A kind of coding method of convolution code may further comprise the steps:
A, default sliding window, wherein said sliding window is made up of judgement output and slave part, with the segmentation of convolution code frame in described each sliding window, make the summation of length of the judgement output of described each sliding window equal the convolution code frame length, default sliding window sign iw, and the total L in the globally-optimal selection path that need search, initialization iw equals 0;
B, make iw add up 1, obtain in the iw sliding window, the state of every state point in per moment, every state point in per moment is with respect to the described convolution code path metric of the zero hour, choose L preferred path tolerance according to described path metric, and store described L preferred path and measure corresponding respectively preferred path, state point with moment of the least significant end of iw sliding window is that starting point begins to recall, in per moment, according to the path values of recalling state point, recall according to optimal path, date back to the moment of the starting end of iw sliding window, obtain the local optimum path of the judgement output of iw sliding window, and with described local optimum path at the state point in the moment of judgement output least significant end for recalling starting point, preferred path value according to the state point of recalling is recalled, obtain other the local preferred path except that the local optimum path, execution in step C;
If C iw less than the total number of sliding window, returns step B, otherwise, execution in step D;
D, according to each sliding window obtained owning administration quality award from the ministry routing footpath, obtain all local preferred paths and the decoding in described each globally-optimal selection path, and the decoding data that the globally-optimal selection path that is obtained is exported carries out error checking and correction, according to check results, the output decode results, wherein said local preferred path comprises the local optimum path, and other the local preferred path except that the local optimum path.
Method of the present invention, being provided with specifically of the sliding window described in the steps A: the length of described judgement output is set, and the length of slave part, wherein the length of slave part specifically is, according to the decoding depth setting of described convolution code.
Method of the present invention, step B is described to choose L preferred path tolerance specifically according to described path metric: if the input of described Convolutional Decoder Assembly is a hard decision, then select less L path metric to measure as described preferred path; If described Convolutional Decoder Assembly input is a soft-decision, then select L bigger path metric, as described preferred path tolerance.
Specifically with a numeric representation with L bit cell, wherein the value of each bit cell is represented each preferred path respectively for method of the present invention, the corresponding respectively preferred path of described L preferred path tolerance.
Preferably, in described step B, in the local optimum path process of the described judgement output that obtains the iw sliding window, if the input of described Convolutional Decoder Assembly is a hard decision, the state point of path metric minimum in the moment of then choosing the least significant end of iw sliding window is that starting point begins to recall; If the input of described Convolutional Decoder Assembly is a soft-decision, the state point of path metric maximum in the moment of then choosing the least significant end of iw sliding window is that starting point begins to recall.
Method of the present invention, the preferred path value of the state point that the basis described in the step B is recalled is recalled, and specifically may further comprise the steps:
B1, according to each preferred path of every state point correspondence in per moment, a default preferred path is recalled table, described preferred path is recalled table, specifically reflects the corresponding relation of each preferred path and described each preferred path value.
B2, in per moment of recalling, obtain current each preferred path of recalling place state point correspondence, and contrast described preferred path and recall table, recall table according to described preferred path, recall and obtain described local preferred path.
Preferably, among the described step B2, if the state point in the per moment before the moment of the starting end that dates back to the iw sliding window, the path values of other all preferred path correspondences that are better than current preferred path of recalling of described state point correspondence is all identical with the current path values of recalling the path, then date back to the moment of the starting end of iw sliding window, stop to recall.
Method of the present invention further may further comprise the steps in step B2:
B21, per moment of recalling, after obtaining current each preferred path value of recalling place state point correspondence, whether preferred path value and optimal path value that the current needs in each preferred path value of comparison are recalled the local preferred path correspondence of obtaining be identical, after for the first time inequality, next moment and later per moment of recalling of recalling, obtain the state of the state point of recalling, and the state of more described state point constantly state is identical at described state point place with the local optimum path, if different, continue to recall; Otherwise, stop to recall, preserve current local preferred path of recalling.
Preferably, behind the current local preferred path of recalling of the described preservation of step B21, further comprise: other identical local preferred paths of the path number with current preferred path of recalling of preserving before giving up.
Preferably, in described step B, further may further comprise the steps:
If B3 iw, then discharges state, the path metric of sliding window sign less than every state point in the per moment in all every other sliding windows of (iw-1) greater than 2, and the memory space in path.
Method of the present invention, the obtaining specifically of the global optimum path described in the step D:, obtain the global optimum path according to the local preferred path of each sliding window that is obtained.
Method of the present invention, the obtaining specifically of other globally-optimal selection path except that the global optimum path described in the step D: according to each local preferred path of preserving, and each local optimum path, obtain described globally-optimal selection path.
Method of the present invention, the decoding data of among the step D globally-optimal selection path that is obtained being exported carries out error checking and correction, and according to check results, the output decode results specifically may further comprise the steps:
D1, a default real variable l, initialization l equals 0;
D2, the decoding data of overall l shortest path output is carried out error checking and correction, if verification is correct, execution in step D4, otherwise, further judge the sum in the globally-optimal selection path whether l searches less than default need, if less than, D3 carried out; Otherwise, execution in step D4;
D3, make l add up 1, return D2;
D4, output decode results finish decoding
As can be seen from the above technical solutions, the present invention has following advantage with respect to prior art:
At first owing in traditional PLVA technology, introduced the technology of sliding window, with the segmentation of convolution code frame in individual sliding window, in each sliding window, obtain local preferred path,, obtain the preferred path of the overall situation again according to the local preferred path register value in each sliding window.For frame length is M, and status number is the convolution code of N, and the sum of the preferred path that need search is L, uses traditional PLVA, and the memory space that need take is respectively: L*N*M state storage space, and 2L*N*M tolerance memory space; And use the inventive method, the length of establishing sliding window is W, the memory space that need take is respectively 2L*N*W state storage space, and 2L*N*W tolerance memory space.Obviously, W＜M, so relative prior art of the present invention, this method had both realized the advantage of PLVA convolution code decoding, can reduce the needed memory space of decoding again, and, because the introducing of sliding window setting technique, the memory space that feasible decoding need take is not subjected to the constraint of convolution code length substantially, and practical application is more strengthened.Especially, when the convolution code frame length was longer, the effect of using the inventive method was more remarkable.
Secondly, the invention allows for a kind of method of how in sliding window, to search other the local preferred paths except that the local optimum path.How this method searches the local optimum route method with respect to what disclose in the document 1 in sliding window, both technically the application facet of sliding window technique had been carried out further expanding and replenishing, more aspect practical application, proposed to be fit to more the solution of practical application.
Description of drawings
Fig. 1 is a LVA decoding architecture schematic diagram;
Fig. 2 is the method schematic diagram of PLVA;
Fig. 3 is the inventive method schematic flow sheet;
Fig. 4 is a kind of sliding window schematic diagram of the present invention;
Fig. 5 is an another kind of sliding window schematic diagram of the present invention;
Fig. 6 be per moment every state point correspondence path register value schematic diagram;
Preferred path when Fig. 7 is L=4 is recalled table;
Fig. 8 is that sliding window selects to influence schematic diagram to sub-optimal path;
Fig. 9 is the schematic flow sheet of embodiment 1 method;
Figure 10 is the schematic flow sheet of recalling local the 2nd shortest path at the iw window.
Embodiment
Core concept of the present invention is, sliding window technique is incorporated among the PLVA, specifically realize: at first by following technical scheme, default sliding window, wherein said sliding window is made up of judgement output and slave part, with the segmentation of convolution code frame in described each sliding window, make the summation of length of the judgement output of described each sliding window equal the convolution code frame length, default sliding window marking variable iw, and the total L in the globally-optimal selection path that need search, initialization iw=0; Then, make iw=iw+1, obtain the state of every state point in the per moment in the iw sliding window, and every state point in per moment is with respect to the described convolution code path metric of the zero hour, choose L preferred path tolerance according to described path metric, and store described L preferred path and measure corresponding respectively preferred path, state point with moment of the least significant end of iw sliding window is that starting point begins to recall, in per moment, according to the path values of recalling state point, recall according to optimal path, date back to the moment of the starting end of iw sliding window, obtain the local optimum path of the judgement output of iw sliding window, and with described local optimum path at the state point in the moment of judgement output least significant end for recalling starting point, preferred path value according to the state point of recalling is recalled, obtain other local preferred path, if iw is less than the total number of sliding window, then continue to make iw add 1, in next sliding window, search corresponding local optimum path and other the local preferred paths except that optimal path; If iw is not less than the total number of sliding window, according to the owning administration quality award from the ministry routing footpath that each sliding window obtained, obtain all local preferred paths and the decoding in described each globally-optimal selection path, and the decoding data that the globally-optimal selection path that is obtained is exported carries out error checking and correction, according to check results, the output decode results, wherein said local preferred path comprises local optimum path and sub-optimal path.
In order to allow those skilled in the art better understand content of the present invention, be further detailed below in conjunction with accompanying drawing and embodiment:
Be illustrated in figure 3 as the schematic flow sheet of the inventive method, as shown, the present invention includes following steps:
Step S301: sliding window is set, the globally-optimal selection total number of paths L that default needs are searched, and window ID variable i w is set, and initialization iw=0.
Fig. 4 is a sliding window schematic diagram of the present invention, and as shown, sliding window W is by judgement output W
R, and slave part Wc form.The length W of sliding window W equals to adjudicate output W
RLength R, and slave part Wc length C sum, i.e. W=R+C.
Judgement output W at first is set
R, this sliding window is decoding judgement output, the size of the length R of this part specifically is provided with according to the decoding system performance of reality, if when the decoding system performance is more excellent, long judgement output can be set, promptly bigger R value; If decoding system poor-performing, memory space hour, can be provided with less R value, make that to adjudicate output shorter.The length judgement output of each sliding window and that equal the convolution code of required decoding.
Next slave part Wc, the setting of slave part Wc, if mainly be because each sliding window all has only the judgement output, and auxiliary output is not set, according to the definition of optimal path and the characteristic of grid chart, the optimal path in each sliding window may not be the part path of optimal path in this sliding window on the global sense so.For example, for a length is the convolution code of M, suppose when the moment (M-1), the path metric minimum of state point A in all state points, if this minimal path tolerance is a, and the path of path metric a correspondence is path 1, and state point A transfers to state point B at moment M, and state point A is ab to the branch metric of state point B, characteristic according to grid chart, also exist another to import branch for state point B, establish the remittance by state point C, the path metric of state point C is c, and state point C is ac to the branch metric of state point B, if ac＞ab, and make (c+ac)＜(a+ab), promptly at moment M, the path of path metric minimum, being the global optimum path, is the path of state point C remittance, rather than the path of state point A remittance.As seen, from convolution code zero hour to the optimal path of (M-1) constantly is not the part in global optimum path, if be starting point from (M-1) constantly, when recalling optimal path according to VA, recalling the local path that obtains is path 1, can not obtain correct optimal path.Therefore, at this situation, be provided with slave part at the judgement output end of each sliding window.Characteristic according to VA: when decoding depth d was enough big, when equaling 4 to 6 times of constraint length of encoder for convolution codes such as, d, all survivor paths can merge in the unit at d constantly with very big probability.Therefore, the length C of slave part is set to enough big, make all survivor paths in moment length C, to merge with very big probability, in all sliding windows that comprise judgement output and slave part, carry out forward calculation, and the moment at the least significant end of slave part is the initial moment, recalls according to optimal path.Setting according to the length C of slave part, recall with very big probability in time judgement sliding window with real global sense on optimal path merge, guarantee in judgement output sliding window, to recall the part that the path that obtains is the optimal path on the global sense.
According to above method to set up to sliding window, can learn, the length that is no more than a sliding window of the very big probability of length that each globally-optimal selection path and global optimum path are separated, that is, if the globally-optimal selection path is separated with the global optimum path in certain sliding window, so very big probability, will be in this sliding window, perhaps, (specifically being in the slave part of another sliding window) merges mutually with the global optimum path in another adjacent with it sliding window.
With the segmentation of convolution code frame in each sliding window, make the summation of length of the judgement output of described each sliding window equal the convolution code frame length, each sliding window on convolution code can be the same or different, type specifically convenient according to decoding and convolution code is provided with, for decoding common in the communication system to whole frame convolution code, as shown in Figure 4, usually set and make other all sliding windows except that the sliding window that is positioned at convolution code frame least significant end the R that equally is each sliding window, the C parameter is the same, and for the sliding window of least significant end, make the length R of the judgement output of this sliding window equal remaining convolution code length, the length C of its slave part equals the redundant code segment length of convolution code end (generally when convolutional encoding, after end-of-encode, terminal each bit value of importing (K-1) bit at coding are the redundant code section of " 0 " more, are used to confirm end-of-encode).
And if desired to certain section of a frame convolution code frame when deciphering, usually adopt sliding window setting shown in Figure 5, the slave part C parameter unanimity of all sliding windows, and for other sliding window except the sliding window of the convolution code section least significant end that is positioned at required decoding, its judgement output R parameter is the same, and specifically equals the length of remaining convolution code part for the judgement output R parameter of the sliding window of the least significant end of the convolution code section that is positioned at required decoding.
Step S302:iw=iw+1 obtains the local optimum path of iw sliding window and other the local preferred paths except that optimal path.
In the moment of convolution code frame order from big to small, the sliding window of moment corresponding minimum is called first sliding window according to sliding window, second sliding window secondly, and the like.
Make iw add 1, iw=1 at first, at first to the 1st sliding window:
In the grid chart in the 1st sliding window, from zero hour of sliding window (zero hour that also promptly needs the convolution code deciphered for first sliding window), forward calculation, for each this state of each state storage constantly, and calculate 2 branch metrics respectively, and respectively the path metric of the state point of previous moment is added branch metric, obtain the path metric of current state point.If required globally-optimal selection path of searching add up to L, then calculate 2L path metric, and preserve less L path metric in this 2L path metric, and the path of this L that preserves path metric correspondence at each state point.According to optimum, suboptimum, the 3rd excellent until excellent order of L, corresponding stored.
Fig. 6 be per moment every state point correspondence path register value schematic diagram, this path register value specifically is each preferred path that is stored in this state point correspondence in the register of path, the total number of paths of supposing required globally-optimal selection path of searching is L, as shown, all there is corresponding path register to store this L preferred path at each each state point constantly, in the register of path, preserve L bit value, every bit value represent respectively to arrive current state point migration each preferred path constantly, such as, the 1st bit is represented the path of optimal path, the 2nd bit is represented the path of sub-optimal path, the 3rd bit is represented the 3rd shortest path, successively, the L bit is represented the path of L shortest path.The value of every bit can be " 0 " or " 1 ", bit value specifically represents to arrive the path of the previous moment of this state point, characteristic according to grid chart in the mobile communication, for each every state point constantly, all there are 2 branches to import, 2 branch's remittance abroads also arranged, therefore, can use bit value " 0 " and " 1 " respectively current state specifically come by that branch transition, promptly current state is that state point by two state points of previous moment shifts and comes.Such as, if bit value is set when " 0 ", the expression current state is that certain state point by previous moment shifts and comes, and bit value represents that then current state is to be come by another state transitions of previous moment when being " 1 ".
Forward calculation is after the moment of the least significant end of sliding window, and the forward calculation in the current sliding window finishes.The state of all state points in the current window, L the preferred path value (being the path register value) of all state points have been preserved this moment.
Retrospective search local optimum path: the moment of choosing the least significant end of the slave part in the sliding window, optional state point is a starting point, path register value according to every state point, recall according to optimal path, date back to the zero hour of this sliding window always, promptly dating back to the zero hour of judgement output sliding window, is needed local optimum path in the path that the judgement output is recalled wherein, preserves the judgement output optimal path of scope constantly.
Search other the local preferred paths except that the local optimum path: choosing the state point of local optimum path in the moment of the least significant end of the judgement output of sliding window that is obtained is starting point, per moment of recalling, according to the path register value of recalling the place state point, the needed local preferred path of retrospective search, merge mutually with the local optimum path if recall in current window always, then date back to till the moment of the starting end of current window, stop to recall; Otherwise, date back to always with local preferred path till merging.
For convenience of the needed L of retrospective search local preferred path, can be according to path register value as shown in Figure 6, set up a preferred path and recall table, preferred path when being illustrated in figure 7 as L=4 is recalled table, as shown, " 1 " expression in the table:, recall according to optimal path in next moment of recalling; " 2 " expression:, recall according to sub-optimal path in next moment of recalling; " 3 " expression:, recall according to the 3rd shortest path in next moment of recalling; " 4 " are illustrated in next moment of recalling, recall according to the 4th shortest path.Therefore this table has been described, when searching each local preferred path, retrospective search each constantly, recall the path register value of place state point constantly according to this, determine in next moment of recalling specifically should recall according to which path in described next path register value constantly.For example:
If the path of current retrospective search is an optimal path, contrast form shown in Figure 7, as seen what the path register value of each state point correspondence of no matter recalling is, constantly all recalls according to optimal path in each constantly next of recalling.When dating back to the moment of current window starting end, get access to the local optimum path of current line window.
If, the path of current retrospective search is sub-optimal path (promptly local the 2nd shortest path), recall table according to preferred path shown in Figure 7, if sub-optimal path is identical with optimal path in the current path values of recalling the place state point, promptly contrast in the form shown in Figure 7, path values is: " 0000 ", " 0001 ", " 0010 ", " 0011 ", " 1111 ", " 1110 ", " 1101 ", " 1100 ", then recall next constantly all according to the path register value of the state point in described next moment, recall according to sub-optimal path; If sub-optimal path and optimal path are inequality in the current path values of recalling the place state point, promptly contrast in the form shown in Figure 7, path values is: " 1000 ", " 1001 ", " 1010 ", " 1011 ", " 0111 ", " 0110 ", " 0101 ", " 0100 ", then, all recall according to the optimal path in described next path register value constantly in next moment of recalling.
The path of current retrospective search is the 3rd shortest path, each moment of recalling, obtain current path register value of recalling the place state point, according to this path register value, contrast form shown in Figure 7, if the optimal path value of path values, sub-optimal path value, the 3rd shortest path value equate that promptly path values is " 0000 ", " 0001 ", " 1110 ", " 1111 ", then, recall according to the 3rd shortest path in described next path register value constantly in next moment of recalling;
If the optimal path value of path values, sub-optimal path value are identical, and the 3rd shortest path value and optimal path value, sub-optimal path value are different, be that path values is: " 0010 ", " 0011 ", " 1101 ", " 1100 ", then, recall according to the optimal path in described next path register value constantly in next moment of recalling;
If the optimal path value of path values, sub-optimal path value difference, one of them is identical and the 3rd shortest path value is with optimal path value, sub-optimal path value, be that path values is: " 010 ", " 0101 ", " 0110 ", " 0111 ", " 1011 ", " 1010 ", " 1001 ", " 1110 ", then, recall according to the sub-optimal path in described next path register value constantly in next moment of recalling;
If, the path of current retrospective search is the 4th shortest path, each moment of recalling, obtain current path register value of recalling the place state point,, contrast form shown in Figure 7 according to this path register value, if the optimal path value of path values, sub-optimal path value, the 3rd shortest path value, the 4th shortest path value all equate, be that path values is " 0000 ", " 1111 ",, recall according to the 4th shortest path in described next path register value constantly then in next moment of recalling;
If the optimal path value of path values, sub-optimal path value, the 3rd shortest path value, the 4th shortest path value have only three identical, one of them and other 3 are different, be that path values is: " 0010 ", " 0100 ", " 0111 ", " 1101 ", " 1011 ", " 1110 ", then, recall according to the 3rd shortest path in described next path register value constantly in next moment of recalling;
If the optimal path value of path values, sub-optimal path value, the 3rd shortest path value, the 4th shortest path value have only two identical, be that path values is: " 0011 ", " 0101 ", " 0110 ", " 1100 ", " 1010 ", " 1001 ", then, recall according to the sub-optimal path in described next path register value constantly in next moment of recalling;
If the optimal path value of path values, sub-optimal path value, the 3rd shortest path value are identical, and the 4th shortest path value and optimal path value, sub-optimal path value, the 3rd shortest path value are different, be that path values is: " 0001 ", " 1110 ", then, recall according to the optimal path in described next path register value constantly in next moment of recalling;
In per moment of recalling, recall table according to preferred path shown in Figure 7, recall respectively and obtain a local preferred path.What deserves to be explained is, in trace-back process, when if the optimal path value of the path values path values corresponding with current local preferred path of recalling is for the first time inequality (such as, if the path of current retrospective search is local sub-optimal path, then the optimal path value of path values is for the first time different with the sub-optimal path value; If the path of current retrospective search is local the 3rd shortest path, then the optimal path value of path values is for the first time different with the 3rd shortest path value.) per moment of then recalling afterwards, obtain the state of recalling the place state point, whether this state is identical at corresponding state constantly with known local preferred path, if it is different, then recall table according to preferred path shown in Figure 7 recalls always, date back to always, recall the state and the state same position of local preferred path of place state point, stop to recall at corresponding state point constantly.
The position of sliding window can exert an influence to recalling of other preferred paths except that optimal path, below is that example describes with the sub-optimal path, and other preferred path and sub-optimal path are in like manner.
Show that as Fig. 8 in the time of in sub-optimal path drops on sliding window, as sliding window 3, obviously, recalling in sliding window 3 of sub-optimal path stops, and is not subjected to the influence of sliding window.But.When sliding window drops on the mid portion of sub-optimal path and optimal path, position as sliding window 4, then in current sliding window, recall according to the method for traditional sliding window, can't recall and complete sub-optimal path, need utilize last sliding window this moment beyond sliding window 4, it is the routing information of sliding window 3, continue to recall,, could correctly obtain sub-optimal path up to till the optimal path of sliding window 3 merges mutually.
In current sliding window, all other all (L-1) individual preferred paths except optimal path are recalled, and after storing each the local preferred path that is obtained, execution in step S303, whether judge current iw less than the sliding window sum, if be not less than, execution in step S304 then; Otherwise, return step S302, continue to make iw to add 1, make iw=2.
Forward calculation in the 2nd sliding window is basic identical with the calculating in first sliding window, different is: in this sliding window, because the moment of the least significant end of the judgement output of the 1st sliding window is the zero hour of the judgement output of second sliding window, the slave part of the 1st sliding window and second sliding window coincide.State, cumulative metric and the path values of all state points of the part that coincides, when the first sliding window forward calculation, obtained, therefore, when the 2nd sliding window is calculated, can carry out double counting, and from the moment of the least significant end of the 1st sliding window be the zero hour of forward calculation, if should constantly be moment i, so, the path metric of all state points of the moment i that obtains during according to iw=1, next branch metric constantly of accumulative total obtains next path metric constantly.And, preserve the path values of L preferred path according to path metric.
After forward calculation finishes, and obtain same method in the 1st sliding window, obtain the local optimum path of the judgement output of the 2nd sliding window.Choose this local optimum path at the state point in the moment of the end of the judgement output of the 2nd sliding window for recalling starting point, with the same method of in the 1st sliding window, obtaining except that optimal path of preferred path, the corresponding preferred path of retrospective search in the 2nd sliding window.What deserves to be explained is that when other preferred paths except that optimal path of retrospective search if the preferred path of being recalled keeps overlapping with optimal path, moment of starting end of then dating back to the judgement output of this sliding window stops in current sliding window; Otherwise date back to till optimal path merges mutually always, and store the current preferred path that obtains with the optimal path path register value of section constantly that is separated, recall identical local preferred path (if the local optimum path in the 2nd local preferred path that promptly in the 2nd sliding window, obtains and this window of the path number with the current local preferred path that obtains that obtains and give up at last sliding window, 2 sliding window inside overlap constantly at all on ground, give up the local sub-optimal path that in the 1st sliding window, obtains so, for other local preferred paths, in like manner handle).If after preferred path of being recalled and optimal path are separated in current sliding window, date back to the moment of the starting end of current sliding window, still not with current sliding window in optimal path merge, continue so in last sliding window, to recall, till the local preferred path of preferred path of recalling and last sliding window merges.In the 2nd sliding window, all preferred paths are recalled finish, and after having upgraded the path register value of described each preferred path, execution in step S303, judge that whether current iw is less than the sliding window sum, if be not less than, execution in step S304 then, otherwise, represent that current sliding window is not the sliding window of least significant end, return step S302, continue to make iw to add 1, make iw=3.
To the acquisition methods in the optimal path of the 3rd sliding window and follow-up other sliding windows and other preferred paths and the 2nd sliding window in like manner.Do not give unnecessary details at this.
What deserves to be explained is, according to top description, as can be seen, when retrospective search removes other preferred paths in global optimum path, exist the preferred path recall in current sliding window, not merge with the local optimum path of current sliding window, and need to proceed to recall in front the sliding window, recall in the sliding window in front with this sliding window in the local optimum path merge mutually possible.Basis is learnt the setting of the length C of the slave part Wc of sliding window again, if a preferred path of recalling merges with optimal path in current sliding window, must merge with optimal path in last sliding window.Therefore, if current sliding window is the iw sliding window, need, and only need store the state of all state points in (iw-1) sliding window, the L of all state points footpath tolerance and with respect to the path values of L path metric, so that the iw sliding window is searched other the local preferred paths uses except that the local optimum path in this sliding window.
Step S303: judge current iw whether less than the sum of sliding window, if be not less than, then expression when sliding window be the sliding window of least significant end, execution in step S304; Otherwise return step S302.
Step S304: according to the owning administration quality award from the ministry routing footpath that each sliding window obtained, obtain all local preferred paths and the decoding in described each globally-optimal selection path, and the decoding data of the globally-optimal selection path that obtained output carried out error checking and correction, and according to check results, the output decode results.
Obviously, because the judgement output of each adjacent sliding window head and the tail are connected at the synchronization point, so local preferred path of each sliding window, according to the combination of order constantly is the global optimum path, and for other each the globally-optimal selection paths except that the global optimum path, can be according to each preferred path and global optimum the path of moment section through being separated, in conjunction with the global optimum path, the path of moment section with each preferred path and global optimum through being separated, replace the path of the corresponding moment section on the global optimum path, just can obtain each the globally-optimal selection path that comprises all moment.
To the decoding data in each globally-optimal selection path, carry out error checking and correction, if the decoding error checking and correction in certain globally-optimal selection path is correct, then export this correct decoding, finish decoding; If it is all incorrect that the decoding data of all paths output is carried out error checking and correction respectively, then export the decode results of decoding failure, finish decoding.
The method of the existing error checking and correction that can adopt specifically has: CRC check, parity check, duplication code verification, constant ratio code verification etc.Specifically be when chnnel coding, when channel transmission data is encoded, in data, insert error check code, make check code send to the channel receiving terminal with data.The channel receiving terminal carries out demodulation to received signal, after the decoding, obtain decoding data, according to chnnel coding being adopted different error checking and correction coded systems, when decoding data is carried out error checking and correction, corresponding error checking and correction mode when adopting with chnnel coding.Such as, if when chnnel coding, employing be the CRC check sign indicating number, when so error checking and correction is carried out in decoding, corresponding employing CRC check mode.
When decoding failure, communication system can be deciphered according to this, carries out corresponding error control strategy such as automatic repeat request, forward error correction or hybrid error correction.
Below will do further to analyze to the inventive method by specific embodiment:
Figure 9 shows that the method flow schematic diagram of present embodiment, as shown, this method may further comprise the steps:
Step S901: default sliding window.
According to the convolution code frame length, the judgement output W of sliding window is set
RLength R, and the length C of the part Wc that coincides with next sliding window, wherein the summation of the judgement output R of each sliding window equals the convolution code frame length.
As shown in Figure 4, for other the sliding window except that last sliding window that is positioned at the convolution code least significant end, judgement output W
RLength R size be R0, the length C size of the part Wc that coincides with next sliding window is C0=6*K, wherein K is the constraint length of convolution coder.For frame length is the convolution code of M, and the number Nw of sliding window is as can be known:
Therefore, for the sliding window that is positioned at the latter end of convolution code: the judgement output W of this sliding window
RLength be: R
Nw=M-(Nw*R0), and for the slave part Wc of this sliding window because this sliding window is positioned at the latter end of convolution code, can it be set to C
Nw=K-1.
Step S902: the current sliding window sign of initialization iw=1, and path number l=1.
If the scope of sliding window sign iw is 0＜iw＜=Nw, initialization makes current sliding window be, is positioned at the 1st sliding window of the opening code section of convolution code.
Step S903: variable j constantly is set, and initialization j=0, and determine the scope of j constantly.
According to the iw sliding window, determine the scope of j constantly, promptly the scope of j is constantly, and minimum constantly be the starting end moment corresponding of the judgement output of current sliding window, and maximum moment j is the least significant end moment corresponding of the slave part of current sliding window.
Step S904:j=j+1 obtains branch metric, the path metric of j constantly, and the path register value of L preferred path.
J adds 1 constantly, obtain the state of all state points of j constantly, and Branch Computed tolerance, and the path metric of the previous state point by calculate importing this state point and branch metric with, obtain path metric, if the globally-optimal selection path that need search add up to L, obviously, need preserve L minimum path metric at every state point, and the path of this L path metric correspondence, because the previous moment in per moment has L path metric, and the pointer in back a period of time in per moment has 2 branch metrics to enter this moment, as seen, all needs to calculate 2L path metric for per moment, in a resulting 2L path metric, choose the storage of less L path metric, and the path values of preserving this L path metric correspondence is stored in the register of path.As shown in Figure 6 every state point correspondence in per moment path register value schematic diagram, in this path register, preserve L bit value, from the 1st bit, the 2nd bit,,, the value of L bit, respectively corresponding optimal path, sub-optimal path,,, the routing information of L shortest path.Value to individual every bit may be that " 0 " also may be " 1 ".Because every state point all existing two may branch import, two possibility branch remittance abroads are also arranged, promptly for every state point, only may come by two state transitions of previous moment.Therefore, if bit value is 0, represent then that previous state point is specifically shifted by previous moment one state point to come; If this bit value is 1, represent then that previous state point is specifically shifted by the other state point of previous moment to come.As seen, if the value of dibit is identical, represent that then two paths are same point at the state point of previous moment; Otherwise two paths are in the state point difference of previous moment.
Step S905: judge that j whether in the determined moment scope of step S903, if in moment scope, returns step S904, otherwise, execution in step S906.
Step S906: the state point of path metric minimum in all state points in the most last moment of getting the iw sliding window carries out the path to recall for recalling starting point, obtains judgement output W
RThe local optimum path.
The state point of path metric minimum in all state points in the most last moment of choosing the iw sliding window is for recalling starting point, path register value according to each state point that forward calculation is obtained, recall according to optimal path, export the judgement output W of iw sliding window
RThe local optimum path, and store the path register value in described local optimum path.
In this step, except the state point of path metric minimum in all state points in the most last moment of choosing the iw sliding window for recalling the starting point, the free position point that can also choose other is starting point, different is: the state point of path metric minimum in all state points in the most last moment of choosing the iw sliding window is for recalling starting point, help recalling as early as possible with sliding window in the local optimum paths merge, help dating back to as early as possible on the local optimum road of iw sliding window.
Step S907: recall (L-1) that obtain other individual local preferred path according to the local optimum path, and upgrade corresponding path register value.
Local optimum path according to the judgement output of the iw sliding window that has obtained, be chosen at optimal path at the state point of least significant end moment corresponding of judgement output for recalling starting point, the path register value of each state point of storing according to forward calculation, contrast preferred path shown in Figure 7 and recall table, recall, in order to increase reliability, in the process of recalling, if preferred path is with after optimal path separates, then according to corresponding path register value, after dating back to preferred path and optimal path overlapping, just stop to recall.
Below be example with the sub-optimal path, specify how to recall and obtain sub-optimal path according to the path register value:
At first, be chosen on the optimal path W
RThe state point in the most last moment be starting point, in starting point constantly, i.e. W
RThe most last moment, obtain the path register value of starting point, and according to the path register value, contrast preferred path shown in Figure 7 and recall table, search and be informed in next that recall and constantly should recall, and in this starting point constantly, recall according to sub-optimal path according to that preferred path.Next moment of recalling, it is the previous moment of starting point, path register value according to current time, according to should recalling according to the preferred path which preferred path is recalled of being known at the path in initial moment register value at current time, and at current time, according to the path register value of current time, contrast preferred path shown in Figure 7 and recall table, be informed in down for the moment, should recall, in like manner for each moment of after this recalling according to that preferred path.Be the schematic flow sheet of recalling local the 2nd shortest path at the iw window as shown in figure 10, as shown, specific as follows:
Step S1001: initialization is t=te constantly.
Initialization constantly t equals moment of least significant end of the judgement output of iw sliding window, establishes this and is te constantly constantly;
Step S1002: obtain the path register value that moment te recalls place state point correspondence; And the contrast preferred path recalls table and is informed in constantly that (te-1) should recall according to that preferred path, obtains the path number that (te-1) constantly constantly should recall; At moment te,, recall according to sub-optimal path according to the path register value.
Contrast preferred path shown in Figure 7 and recall table, as seen, at moment te, if the path register value is: " 0000 ", " 0001 ", " 0010 ", " 0011 ", " 1111 ", " 1110 ", " 1101 ", " 1100 ", then in the moment (te-1), all, recall according to sub-optimal path according to the path register value of the state point of (te-1) constantly; Otherwise,,, recall according to optimal path according to the path register value of the state point of (te-1) constantly at constantly (te-1).At moment te,, recall simultaneously according to sub-optimal path according to the path register value of moment te.
Step S1003: according to the path register value, whether the sub-optimal path value is different with the optimal path value in the register value of differentiation path, if different execution in step S1004 '; Otherwise, execution in step S1004.
Step S1004: make moment t equal t and subtract 1, execution in step S1005.
Step S1005: obtain the path register value that moment t recalls place state point correspondence, execution in step S1006 '.
Step S1006: the contrast preferred path is recalled table and is informed in constantly that (t-1) should recall according to that preferred path, obtain the path number that constantly (t-1) should recall constantly, and at moment t, according to the path register value, according to obtain at constantly (t+1) moment t according to preferred path recall, and return step S1007.
Contrast preferred path shown in Figure 7 and recall table, as seen, at moment t, if the path register value is: " 0000 ", " 0001 ", " 0010 ", " 0011 ", " 1111 ", " 1110 ", " 1101 ", " 1100 ", then in the moment (t-1), all, recall according to sub-optimal path according to the path register value of the state point of (t-1) constantly; Otherwise,,, recall according to optimal path according to the path register value of the state point of (t-1) constantly at constantly (t-1).During simultaneously according to (t+1) constantly, obtained moment t should according to concrete preferred path recall.
Step S1007: judge t constantly whether greater than moment of the starting end of iw window, if greater than, then return step S1003; Otherwise, execution in step S10010.
Step S1004 ': make moment t equal t and subtract 1, execution in step S1005 '.
Step S1005 ': obtain the path register value that moment t recalls place state point correspondence, execution in step S1006 '.
Step S1006 ': the contrast preferred path is recalled table and is informed in constantly that (t-1) should recall according to that preferred path, obtain the path number that constantly (t-1) should recall constantly, and at moment t, according to the path register value, according to obtain at constantly (t+1) moment t according to preferred path recall execution in step S1007.
Contrast preferred path shown in Figure 7 and recall table, as seen, at moment t, if the path register value is: " 0000 ", " 0001 ", " 0010 ", " 0011 ", " 1111 ", " 1110 ", " 1101 ", " 1100 ", then in the moment (t-1), all, recall according to sub-optimal path according to the path register value of the state point of (t-1) constantly; Otherwise,,, recall according to optimal path according to the path register value of the state point of (t-1) constantly at constantly (t-1).During simultaneously according to (t+1) constantly, obtained moment t should according to concrete preferred path recall.
Step S1007 ': store local sub-optimal path path.
Step S1008 ': whether the state that judgement moment t recalls the place state point is identical at the state of the state point of moment t with local preferred path, if identical execution, execution in step S1008; Otherwise, return step S1004 '.
If date back to the starting end moment corresponding of iw window, recall also and do not overlap with optimal path, enter (iw-1) window so, continue to recall, till dating back to local optimum path with (iw-1) sliding window and merging according to the path register value in per moment.Setting according to the slave part length C in the sliding window, as seen all preferred paths and optimal path merge in C the moment with very big probability, the length that promptly makes each preferred path and optimal path be separated is not more than a sliding window length, so recalling at the iw sliding window, if in the iw sliding window, do not merge mutually, recall one so and in (iw-1) window, coincide surely with optimal path with optimal path.
If the local sub-optimal path of recalling has been separated with the local preferred path of this window in the iw window, give up the local sub-optimal path that obtains at sliding window before so, store the current local sub-optimal path that obtains of recalling.Only may in a continuous moment section, be separated according to each preferred path and optimal path, therefore, only storage and the optimal path path of section constantly that is separated is gone through at last all over all sliding windows and is searched the back in conjunction with the global optimum path, obtains the global path of this part preferred path.
Step S1009 ': the sub-optimal path that obtains before giving up.
Step S10010: stop to recall.
The 3rd excellent, the 4th shortest path and recalling with ground of other preferred path have the path to recall in like manner, do not give unnecessary details at this.
Step S908:iw=iw+1.
Make iw add 1, upgrade sliding window sign iw.
Step S909: if iw is not less than the sum of sliding window, execution in step S910; Otherwise, return step S903.
Step S910: obtain the excellent preferred path of overall l according to the path register, and the decoding data of l shortest path output is carried out error checking and correction.
So far, according to the path register information of being stored, can get access to the globally-optimal selection path that all need be searched.Recall overall l preferred path according to register, obtain the decoding data of overall l preferred path correspondence, and decoding data is carried out error checking and correction.
Step S911: error checking and correction result judgement.
If error checking and correction is correct, execution in step S914; Otherwise, execution in step S912.
Step S912:l=l+1.
Make path number add 1, upgrade path number.
Step S913: judge that whether l is greater than L.
If l is not more than the total L in the globally-optimal selection path that default need search, execution in step S914, otherwise return step S910.
Step S914: output decode results.
If among the step S911, error checking and correction is correct, and Shu Chu decode results is the decoding data of overall l shortest path so, finishes decoding.Otherwise the decode results of output is: decoding failure, finish decoding.
More than a kind of coding method of convolution code provided by the present invention is described in detail, used specific case herein principle of the present invention and execution mode are set forth, the explanation of above embodiment just is used for helping to understand method of the present invention and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to thought of the present invention, the part that all can change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.
Claims (13)
1, a kind of coding method of convolution code is characterized in that, may further comprise the steps:
A, default sliding window, wherein said sliding window is made up of judgement output and slave part, with the segmentation of convolution code frame in described each sliding window, make the summation of length of the judgement output of described each sliding window equal the convolution code frame length, default sliding window sign iw, and the total L in the globally-optimal selection path that need search, initialization iw equals 0;
B, make iw add up 1, obtain in the iw sliding window, the state of every state point in per moment, every state point in per moment is with respect to the described convolution code path metric of the zero hour, choose L preferred path tolerance according to described path metric, and store described L preferred path and measure corresponding respectively preferred path, state point with moment of the least significant end of iw sliding window is that starting point begins to recall, in per moment, according to the path values of recalling state point, recall according to optimal path, date back to the moment of the starting end of iw sliding window, obtain the local optimum path of the judgement output of iw sliding window, and with described local optimum path at the state point in the moment of judgement output least significant end for recalling starting point, preferred path value according to the state point of recalling is recalled, obtain other the local preferred path except that the local optimum path, execution in step C;
If C iw less than the total number of sliding window, returns step B, otherwise, execution in step D;
D, according to each sliding window obtained owning administration quality award from the ministry routing footpath, obtain all local preferred paths and the decoding in described each globally-optimal selection path, and the decoding data that the globally-optimal selection path that is obtained is exported carries out error checking and correction, according to check results, the output decode results, wherein said local preferred path comprises the local optimum path, and other the local preferred path except that the local optimum path.
2, coding method of convolution code according to claim 1, it is characterized in that, being provided with specifically of sliding window described in the steps A: the length that described judgement output is set, and the length of slave part, wherein the length of slave part specifically is the decoding depth setting according to described convolution code.
3, coding method of convolution code according to claim 1, it is characterized in that, step B is described to choose L preferred path tolerance specifically according to described path metric: if the input of described Convolutional Decoder Assembly is a hard decision, then select less L path metric to measure as described preferred path; If described Convolutional Decoder Assembly input is a soft-decision, then select L bigger path metric, as described preferred path tolerance.
4, coding method of convolution code according to claim 3 is characterized in that, specifically with a numeric representation with L bit cell, wherein the value of each bit cell is represented each preferred path respectively to the corresponding respectively preferred path of described L preferred path tolerance.
5, coding method of convolution code according to claim 4, it is characterized in that, in described step B, the local optimum path process of the described judgement output that obtains the iw sliding window in, if the input of described Convolutional Decoder Assembly is a hard decision, the state point of path metric minimum in the moment of then choosing the least significant end of iw sliding window is that starting point begins to recall; If the input of described Convolutional Decoder Assembly is a soft-decision, the state point of path metric maximum in the moment of then choosing the least significant end of iw sliding window is that starting point begins to recall.
6, coding method of convolution code according to claim 5 is characterized in that, the preferred path value of the state point that the basis described in the step B is recalled is recalled, and specifically may further comprise the steps:
B1, according to each preferred path of every state point correspondence in per moment, a default preferred path is recalled table, described preferred path is recalled table, specifically reflects the corresponding relation of each preferred path and described each preferred path value.
B2, in per moment of recalling, obtain current each preferred path of recalling place state point correspondence, and contrast described preferred path and recall table, recall table according to described preferred path, recall and obtain described local preferred path.
7, coding method of convolution code according to claim 6, it is characterized in that: among the described step B2, if the state point in the per moment before the moment of the starting end that dates back to the iw sliding window, the path values of other all preferred path correspondences that are better than current preferred path of recalling of described state point correspondence is all identical with the current path values of recalling the path, then date back to the moment of the starting end of iw sliding window, stop to recall.
8, coding method of convolution code according to claim 6 is characterized in that, further may further comprise the steps in step B2:
B21, per moment of recalling, after obtaining current each preferred path value of recalling place state point correspondence, whether preferred path value and optimal path value that the current needs in each preferred path value of comparison are recalled the local preferred path correspondence of obtaining be identical, after for the first time inequality, next moment and later per moment of recalling of recalling, obtain the state of the state point of recalling, and the state of more described state point constantly state is identical at described state point place with the local optimum path, if different, continue to recall; Otherwise, stop to recall, preserve current local preferred path of recalling.
9, coding method of convolution code according to claim 8 is characterized in that, behind the current local preferred path of recalling of the described preservation of step B21, further comprises:
Other identical local preferred paths of the path number with current preferred path of recalling of preserving before giving up.
10, coding method of convolution code according to claim 9 is characterized in that, further may further comprise the steps in described step B:
If B3 iw, then discharges state, the path metric of sliding window sign less than every state point in the per moment in all sliding windows of (iw-1) greater than 2, and the memory space in path.
11, coding method of convolution code according to claim 1 is characterized in that, the obtaining specifically of the global optimum path described in the step D: according to the local preferred path of each sliding window that is obtained, obtain the global optimum path.
12, according to one of the wherein any described coding method of convolution code of claim 1 to 11, it is characterized in that, obtaining specifically of other globally-optimal selection path except that the global optimum path described in the step D: according to each local preferred path of preserving, and each local optimum path, obtain described globally-optimal selection path.
13, coding method of convolution code according to claim 12 is characterized in that, the decoding data of among the step D globally-optimal selection path that is obtained being exported carries out error checking and correction, and according to check results, the output decode results specifically may further comprise the steps:
D1, a default real variable l, initialization l equals 0;
D2, the decoding data of overall l shortest path output is carried out error checking and correction, if verification is correct, execution in step D4, otherwise, further judge the sum in the globally-optimal selection path whether l searches less than default need, if less than, D3 carried out; Otherwise, execution in step D4;
D3, make l add up 1, return D2;
D4, output decode results finish decoding.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100874538A CN100536351C (en) | 2006-06-08 | 2006-06-08 | A coding method of convolution code |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100874538A CN100536351C (en) | 2006-06-08 | 2006-06-08 | A coding method of convolution code |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1968023A true CN1968023A (en) | 2007-05-23 |
| CN100536351C CN100536351C (en) | 2009-09-02 |
Family
ID=38076609
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100874538A Active CN100536351C (en) | 2006-06-08 | 2006-06-08 | A coding method of convolution code |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100536351C (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101997553B (en) * | 2009-08-13 | 2013-03-20 | 中兴通讯股份有限公司 | Method and device for decoding convolution code |
| CN105335747A (en) * | 2014-08-04 | 2016-02-17 | 联想(北京)有限公司 | Data processing method and electronic equipment |
| CN107919937A (en) * | 2016-10-10 | 2018-04-17 | 深圳超级数据链技术有限公司 | Interpretation method, device and modulation-demo-demodulation method and system based on overlapping multiplexing |
-
2006
- 2006-06-08 CN CNB2006100874538A patent/CN100536351C/en active Active
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101997553B (en) * | 2009-08-13 | 2013-03-20 | 中兴通讯股份有限公司 | Method and device for decoding convolution code |
| CN105335747A (en) * | 2014-08-04 | 2016-02-17 | 联想(北京)有限公司 | Data processing method and electronic equipment |
| CN105335747B (en) * | 2014-08-04 | 2019-03-29 | 联想(北京)有限公司 | A kind of data processing method and electronic equipment |
| CN107919937A (en) * | 2016-10-10 | 2018-04-17 | 深圳超级数据链技术有限公司 | Interpretation method, device and modulation-demo-demodulation method and system based on overlapping multiplexing |
| CN107919937B (en) * | 2016-10-10 | 2020-10-27 | 深圳光启合众科技有限公司 | Decoding method and device based on overlapping multiplexing and modulation and demodulation method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100536351C (en) | 2009-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1315264C (en) | Decoding device and decoding method | |
| US7587005B2 (en) | Exploiting known padding data to improve block decode success rate | |
| CN101431388B (en) | Method and apparatus of decoding encoded data frame | |
| EP1841116A3 (en) | Decoding method for tail-biting convolutional codes using a search-depth Viterbi algorithm | |
| EP1628405A2 (en) | Method and system for decoding video, voice, and speech data using redundancy | |
| CN101079642A (en) | Method for decoding error correcting code, its program and its device | |
| EP2392074A1 (en) | Encoding and decoding methods for expurgated convolutional codes and convolutional turbo codes | |
| CN103117753A (en) | Tail-biting convolutional decoding apparatus and decoding method | |
| CN1287542C (en) | Combined equalization and decoding method | |
| CN1731686A (en) | VCP and ACS circuit for VCP | |
| CN1175825A (en) | Trace-back method and apparatus for use in viterbi decoder | |
| CN1302624C (en) | Decoder for trellis-based channel encoding | |
| CN1126314C (en) | Transmission system with adaptive channel encoder and decoder | |
| CN1777256A (en) | Enhanced VSB viterbi decoder | |
| US7644346B2 (en) | Format detection | |
| CN1968023A (en) | A coding method of convolution code | |
| US8055986B2 (en) | Viterbi decoder and method thereof | |
| CN101425869A (en) | Decoding method and apparatus | |
| CN1968024A (en) | A coding method of convolution code | |
| JP2008118327A (en) | Viterbi decoding method | |
| CN1147169C (en) | Decoding method and decoder for Turbo code | |
| CN1741614A (en) | Method and system for decoding video, voice, and speech data using redundancy | |
| CN1582555A (en) | Segmental deinterlacing | |
| CN1129257C (en) | Maximum-likelihood decode method f serial backtracking and decoder using said method | |
| CN1172448C (en) | Method for decoding data signal |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211227 Address after: 450046 Floor 9, building 1, Zhengshang Boya Plaza, Longzihu wisdom Island, Zhengdong New Area, Zhengzhou City, Henan Province Patentee after: Super fusion Digital Technology Co.,Ltd. Address before: 518129 Bantian HUAWEI headquarters office building, Longgang District, Guangdong, Shenzhen Patentee before: HUAWEI TECHNOLOGIES Co.,Ltd. |
|
| TR01 | Transfer of patent right |