CN107294544A - Forward error correction, device and the decoder of convolutional code - Google Patents

Forward error correction, device and the decoder of convolutional code Download PDF

Info

Publication number
CN107294544A
CN107294544A CN201710500396.XA CN201710500396A CN107294544A CN 107294544 A CN107294544 A CN 107294544A CN 201710500396 A CN201710500396 A CN 201710500396A CN 107294544 A CN107294544 A CN 107294544A
Authority
CN
China
Prior art keywords
sequence
convolutional code
smooth
searched
error correction
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
Application number
CN201710500396.XA
Other languages
Chinese (zh)
Other versions
CN107294544B (en
Inventor
王继深
王苑熙
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Terry (beijing) Mdt Infotech Ltd
Original Assignee
Terry (beijing) Mdt Infotech Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Terry (beijing) Mdt Infotech Ltd filed Critical Terry (beijing) Mdt Infotech Ltd
Priority to CN201710500396.XA priority Critical patent/CN107294544B/en
Publication of CN107294544A publication Critical patent/CN107294544A/en
Application granted granted Critical
Publication of CN107294544B publication Critical patent/CN107294544B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/25—Error detection or forward error correction by signal space coding, i.e. adding redundancy in the signal constellation, e.g. Trellis Coded Modulation [TCM]
    • H03M13/256—Error detection or forward error correction by signal space coding, i.e. adding redundancy in the signal constellation, e.g. Trellis Coded Modulation [TCM] with trellis coding, e.g. with convolutional codes and TCM

Abstract

The invention provides a kind of forward error correction of convolutional code, device and decoder, including：The sequence to be searched in convolutional code sequence is extracted, the smooth Penetrating air-flow of directed euler graph rule is met according to the generator matrix of convolutional code, detects whether sequence to be searched can accurately decode；If so, treating search sequence enters row decoding processing, generate normal solution convolutional code sequence using obtained coding sequence and cover by the corresponding convolutional code of convolutional code sequence after noise jamming；Separation subsequent interference sequence is gone forward side by side row decoding processing in strong convergence characteristic and smooth Penetrating air-flow based on convolutional code sequence, the convolutional code after capped；It is based on mathematical thought, uses convolutional code of second mechanism to reception in the way of error correction after first decoding to enter row decoding processing, decoding speed is fast and the degree of accuracy is high；And also the first mechanism of startup supervises the second mechanism to maintain on correct decoding path in real time, is effectively reduced the problem of the second mechanism steps into erroneous path, reliably solves the problems, such as threshold effect.

Description

Forward error correction, device and the decoder of convolutional code
Technical field
The present invention relates to communication decoding technique field, a kind of forward error correction for the anti-interference convolutional code of high speed is specifically designed Method and device.
Background technology
During information transmission and processing, distorted signals and noise jamming are there is, how from distorted signals and to make an uproar Primary signal is obtained in sound pollution, is always engineering circles and academia very concern.Since 1990s, letter In breath field, mobile communication, deep space communication, satellite navigation, remote control, remote sensing etc. are developed rapidly.Comparatively, it is closely related therewith FEC (Forward Error Correction, forward error correction) technology, then be chronically at technology development bottleneck.
FEC belongs to the high reliability communication technology, is based on Reliability codes theory, and how research utilizes anti-interference coding Regularity and relevance, solve optimum reception the problem of.FEC uses one-way transmission channel, and information source end is responsible for carrying out information Anti-interference coding processing, stay of two nights end then performs jamproof decoding treatment process, corrects the mistake that transmitting procedure is produced, to the greatest extent maximum The reliability and real-time of information may be improved.Therefore, FEC is applicable not only to general communications applications, be more suitable for deep space communication, It is same to broadcast the special occasions such as communication.Another reason for FEC is taken seriously is that its decoding algorithm is same in Smart-Its field With important influence.
In modern communicationses and digital processing field, the folding coding algorithm that Viterbi professor proposes is used till today. The basic operation process of the decoder is related to addition, compares and selects (english abbreviation is ACS) three steps, and it decodes process Each step is required for repeating an ACS computing, finally therefrom selects the minimum path of an error code by backtracking again, avoids Error code larger path, thus the algorithm can not resist strong jamming.All there is this " keep away in current various probability decoding algorithms Weight just it is light " phenomenon.In informationized society, environmental disturbances are increasingly serious with artificial disturbance, the algorithm for owning " shiring the heavy work aning choose the light " All it is difficult to referred to as optimal algorithm.At present, the means for improving decoding performance only have the status number M of increase convolution coder, and are increasing While big status number M the length of metric memory and path memory can be made to double, while ACS operands can also be made suddenly Increase.Therefore, above-mentioned decoder is just considerably complicated for the coding of constraint length length.
In addition, when signal to noise ratio is very low, the coding gain of above-mentioned decoder is changed into negative value, it can be produced in the region without error code Raw substantial amounts of decoding error, forms noise amplification (i.e. threshold effect) so that be inferior to information source using the performance after error correcting code End does not encode the process of directly transmission primary signal.
Inventor has found that the existing convolution decoder based on maximum likelihood algorithm has raising decoding performance under study for action Means it is complicated and the problem of there is threshold effect in itself for the coding process of constraint length length, for the problem, mesh It is preceding not yet to propose preferable settling mode.
The content of the invention
In view of this, the purpose of the embodiment of the present invention is to provide a kind of forward error correction of convolutional code, device and translated Code device, thresholding effect that is easy and effective to the coding process of constraint length length and can reliably solving existing decoder presence Answer problem.
In a first aspect, the embodiments of the invention provide a kind of forward error correction of convolutional code, including：
Receive information source end send by the convolutional code sequence after noise jamming；The convolutional code sequence includes multiple convolution Word, the convolution word is that the information source end is generated according to generator matrix and it carries sequence of cells information and the interference being superimposed Sequence；
Extract the sequence to be searched in the convolutional code sequence；
Meet the smooth Penetrating air-flow of directed euler graph rule according to the corresponding generator matrix of convolutional code, wait to search described in detection Whether Suo Xulie can accurately decode；The smooth Penetrating air-flow is that the convolutional code has continuity and the continuity meets The constraint rule of generator matrix setting；
If so, treating search sequence enters row decoding processing, obtained coding sequence is handled using decoding and generates positive deconvolution Code sequence, and utilize described in normal solution convolutional code sequential covering by the corresponding convolutional code of convolutional code sequence after noise jamming；
Directed euler graph is met according to the strong convergence characteristic of convolutional code sequence and the corresponding generator matrix of the convolutional code Subsequent interference sequence is separated in the smooth Penetrating air-flow of rule, the convolutional code after being capped；
Enter row decoding processing to the subsequent interference sequence, obtain maximum-likelihood decoding sequence.
With reference in a first aspect, the embodiments of the invention provide the possible embodiment of the first of first aspect, wherein, root Meet the smooth Penetrating air-flow of directed euler graph rule according to the corresponding generator matrix of convolutional code, whether detect the sequence to be searched Can accurately it decode, including：
The sequence to be searched is divided into multiple deconvolution time slot groups, multiple deconvolution time slot groups is met following first Impose a condition；The sequence to be searched includes the bit of the first setting digit；Described first impose a condition for：Any two is adjacent The deconvolution time slot group include the second overlapping bit of setting digit；
Multiple deconvolution time slot groups using deconvolution formula respectively to division carry out deconvolution calculating, generate and described anti- The corresponding multiple coding sequences of convolution time slot group；Each described coding sequence includes multiple warp cells；
Detect whether multiple warp cells overlapping in arbitrary neighborhood coding sequence in multiple coding sequences are equal；
If so, determining that the sequence to be searched can be decoded accurately.
With reference in a first aspect, the embodiments of the invention provide the possible embodiment of second of first aspect, wherein, root Meet the smooth of directed euler graph rule according to the strong convergence characteristic and the corresponding generator matrix of the convolutional code of convolutional code sequence Subsequent interference sequence is separated in Penetrating air-flow, the convolutional code after being capped, including：
According to the maximum of the generator matrix for generating convolutional code freely away from building for convolutional code sequence described in separating treatment Multiple multistage smooth types of row；The smooth type includes：Unrelated bit to for the related of convolutional code sequence described in separating treatment Bit；The related bits include the characteristic parameter for being used to detect the convolutional code sequence：Each multistage smooth type includes The different and different related bits detection of the related bits the characteristic parameter it is different；The related bits are The point in any a line that the Euler diagram includes；Outside any a line that the unrelated bit then includes for the Euler diagram Point；
According to multiple multistage smooth types, the subsequent interference sequence and the cell sequence are separated from the convolutional code sequence Column information.
With reference to second of possible embodiment of first aspect, the embodiments of the invention provide the third of first aspect Possible embodiment, wherein, according to the maximum of the generator matrix for generating convolutional code freely away from building for separating treatment Multiple multistage smooth types of the convolutional code sequence, including：
According to the maximum of the corresponding generator matrix of the convolutional code freely away from the most short multistage smooth type of structure；
According to the most short multistage smooth type, multiple multistage smooth types, the exponent number of each multistage smooth type are built It is all higher than the exponent number of the most short multistage smooth type.
With reference to the third possible embodiment of first aspect, the embodiments of the invention provide the 4th of first aspect kind Possible embodiment, wherein, the convolutional code is that information source end is generated according to the generator matrix of [3,1,8] nonsystematic convolutional code 's；
The most short multistage smooth type is [3375], and [3375] are quadravalence smooth type and corresponding related bits and unrelated ratio Spy is followed successively by：110、110、111、101；The multistage smooth type at least includes：Nine rank smooth types [614000023] and ten quadravalences Smooth type [51324000000556].
With reference to second of possible embodiment of first aspect, the embodiments of the invention provide the 5th of first aspect kind Possible embodiment, wherein, accorded with according to the corresponding generator matrix of the strong convergence characteristic and the convolutional code of convolutional code sequence The smooth Penetrating air-flow of directed euler graph rule is closed, in the convolutional code after being capped before separation subsequent interference sequence, Also include：
Extract the sequence to be fractionated in the convolutional code sequence；The sequence to be fractionated includes the ratio of the 3rd setting digit It is special；
Classification processing is carried out to the sequence to be fractionated according to the characteristic parameter of convolutional code sequence, multiple multistage sequences are obtained Row；Each described multistage sequence corresponds to the interference pattern of the first setting number；The interference pattern character pair parameter tool There is general character；The characteristic parameter at least includes：Warp equivalence value and error correction equivalence value；The warp equivalence value is to be treated point to described The sequence of cells that level sequence progress warp processing is obtained is identical.
With reference to second of possible embodiment of first aspect, the embodiments of the invention provide the 6th of first aspect kind Possible embodiment, wherein, row decoding processing is entered to the interference sequence, obtains after maximum-likelihood decoding sequence, also wraps Include：
Convolutional encoding processing is carried out to the obtained maximum-likelihood decoding sequence, obtains covering convolutional code sequence；
Using the covering convolutional code sequential covering by the convolutional code sequence after noise jamming.
Second aspect, the embodiment of the present invention additionally provides a kind of fec arrangement of convolutional code, including：
Receiving module, for receive information source end transmission by the convolutional code sequence after noise jamming；The convolutional code sequence Including multiple convolution words, the convolution word is that the information source end is generated according to generator matrix and it carries sequence of cells information And interference sequence；
Abstraction module, for extracting the sequence to be searched in the convolutional code sequence；
Detection module, for meeting the smooth through spy of directed euler graph rule according to the corresponding generator matrix of convolutional code Property, detect whether the sequence to be searched can accurately decode；The smooth Penetrating air-flow is that the convolutional code has continuity And the continuity meets the constraint rule of generator matrix setting；
Overlay module, for when the sequence to be searched can be decoded accurately, row decoding to be entered to the sequence to be searched Processing, the normal solution convolutional code sequence that obtained coding sequence generation is obtained is handled using decoding, and utilize normal solution convolutional code sequence Corresponding normal solution convolutional code covers the corresponding convolutional code of convolutional code sequence by after noise jamming；
Separation module, for the corresponding generator matrix symbol of strong convergence characteristic and the convolutional code according to convolutional code sequence The smooth Penetrating air-flow of directed euler graph rule is closed, subsequent interference sequence is separated in the convolutional code after being capped；
Processing module is decoded, for entering row decoding processing to the subsequent interference sequence, maximum-likelihood decoding sequence is obtained.
With reference to second aspect, the embodiments of the invention provide the possible embodiment of the first of second aspect, wherein, institute Detection module is stated, including：
Division unit, for the sequence to be searched to be divided into multiple deconvolution time slot groups, makes multiple deconvolution time slots Group meets following first and imposed a condition；The sequence to be searched includes the bit of the first setting digit；Described first imposes a condition For：The adjacent deconvolution time slot group of any two includes the overlapping bit of the second setting digit；
Generation unit, based on carrying out deconvolution to multiple deconvolution time slot groups of division respectively using deconvolution formula Calculate, generate multiple coding sequences corresponding with the deconvolution time slot group；Each described coding sequence includes multiple warps Cell；
Detection unit, for detecting that multiple warp cells overlapping in arbitrary neighborhood coding sequence in multiple coding sequences are It is no equal；
Determining unit, for multiple warps letter overlapping in arbitrary neighborhood coding sequence in detecting multiple coding sequences When member is equal, determine that the sequence to be searched can be decoded accurately.
The third aspect, the embodiment of the present invention additionally provides a kind of decoder, and decoder is used to perform any one of first aspect Convolutional code forward error correction.
Forward error correction, device and the decoder of convolutional code provided in an embodiment of the present invention, are extracted using the first mechanism Sequence to be searched in convolutional code sequence, and smoothly passing through for directed euler graph rule is met according to the corresponding generator matrix of convolutional code Characteristic is worn, detects whether sequence to be searched can accurately decode；If so, treating search sequence enters row decoding processing, using obtaining The covering of normal solution convolutional code by the corresponding convolutional code of convolutional code sequence after noise jamming, convolution is then based on using the second mechanism The strong convergence characteristic and the corresponding generator matrix of convolutional code of code sequence meet the smooth Penetrating air-flow of directed euler graph rule, from Subsequent interference sequence is separated in convolutional code after capped, and enters row decoding processing to the subsequent interference sequence, maximum is obtained seemingly Right coding sequence, with the decoder of the prior art based on maximum likelihood folding coding algorithm, improves the means of decoding performance The problem of there is threshold effect in itself for the coding process complexity and decoder of constraint length length is compared, and it is based on mathematics Thought, uses transmission convolutional code of second mechanism to information source end in the way of error correction after first decoding to enter row decoding processing, decoding speed Degree is fast and the degree of accuracy is high；And also start the first mechanism in real time to be monitored the second mechanism, supervise the second mechanism to maintain just On true decoding path, the problem of the second mechanism steps into erroneous path is efficiently solved, reliably solves and only starts one The threshold effect problem of mechanism.
To enable the above objects, features and advantages of the present invention to become apparent, preferred embodiment cited below particularly, and coordinate Appended accompanying drawing, is described in detail below.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be attached to what is used required in embodiment Figure is briefly described, it will be appreciated that the following drawings illustrate only certain embodiments of the present invention, therefore is not construed as pair The restriction of scope, for those of ordinary skill in the art, on the premise of not paying creative work, can also be according to this A little accompanying drawings obtain other related accompanying drawings.Fig. 1 shows a kind of forward error correction side for convolutional code that the embodiment of the present invention is provided The flow chart of method；Fig. 2 shows the flow chart of the forward error correction for another convolutional code that the embodiment of the present invention is provided；Figure 3 show the flow chart of the forward error correction for another convolutional code that the embodiment of the present invention is provided；Fig. 4 shows the present invention The flow chart of the forward error correction for another convolutional code that embodiment is provided；Fig. 5 shows that the embodiment of the present invention is provided Another convolutional code forward error correction flow chart；Fig. 6 shows another convolution that the embodiment of the present invention is provided The flow chart of the forward error correction of code；Fig. 7 shows a kind of forward error correction dress for convolutional code that the embodiment of the present invention is provided The flow chart put；Fig. 8 shows the decoder of [3,1,8] nonsystematic convolutional code at the stay of two nights end that the embodiment of the present invention is provided Structured flowchart.
Icon：11st, receiving module；12nd, the first abstraction module；13rd, detection module；14th, the first overlay module；15th, separate Module；16th, processing module is decoded.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention Middle accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is only It is a part of embodiment of the invention, rather than whole embodiments.The present invention being generally described and illustrated herein in the accompanying drawings is real Applying the component of example can be arranged and be designed with a variety of configurations.Therefore, it is of the invention to what is provided in the accompanying drawings below The detailed description of embodiment is not intended to limit the scope of claimed invention, but is merely representative of the selected reality of the present invention Apply example.Based on embodiments of the invention, the institute that those skilled in the art are obtained on the premise of creative work is not made There is other embodiment, belong to the scope of protection of the invention.
In modern communicationses and digital processing field, the contribution of A.J.Viterbi professors is well-known.He in The maximum likelihood folding coding algorithm proposed before more than 40 years is used till today (based on the technology as FEC typical decoding technique Decoding implement body is referred to be described in detail in the introduction).
But, the above-mentioned decoder based on maximum likelihood folding coding algorithm is present at the coding for constraint length length Reason process is complicated and the problem of there is threshold effect in itself.
Based on above mentioned problem, most representational convolutional code is used as research pair in selection anti-interference coding of the embodiment of the present invention As.The meaning of the decoding technique of Selecting research convolutional code is as follows：First, the digital honeybee that U.S. Qualcomm (high pass) company proposes It is IS-95 standards that nest spread spectrum code division multiple access system (CDMA) is received by telecommunications industry association (TIA).It is anti-that IS-95 standards are selected It is convolutional code to disturb coding.Second, outside mobile communication, deep space communication, (such as supercomputing in go-go electronic technology Machine, speech recognition, neuroid and high speed optical communication etc.) also use forward error correction technique, it be known as by academia be One of information technology that 21 century most attractes attention.3rd, 21 century will realize the target of personal communication, it is necessary to rely on radio communication and Mobile communication technology.2016, global smart phone user reached more than 2,000,000,000, because CDMA is a self-interference system System, FEC is to improve the validated user density of cdma system, reduce one of radiation, the key technology of reduction power consumption.In a word, in state On border, FEC is widely adopted as a standard technique.
(embodiment of the present invention is provided forward error correction, device and the decoder of convolutional code provided in an embodiment of the present invention Convolutional code forward error correction abbreviation W algorithms), propose new knowledge, new viewpoint in all many-sides of FEC, hence it is evident that improve Decode error correcting capability.
Before explanation W algorithms, quote from first《Mathematical thought at all times》Two viewpoints of one book：First, " it should not recognize to appoint What thing is genuine, unless it understands theoretically clearly arrives unquestionable degree ".Second, " set up true in all fields The method of reason is mathematical method ".Because mathematics always based on tight proof, pursues hardheaded conclusion.
Based on this 2 points, W algorithms form the decoding algorithm of the FEC based on mathematical method, and W algorithms at least have two and dashed forward Go out feature：First, W algorithm propose that " forward error correction technique of convolutional code belongs to decisive phenomenon, thus there is accurate decoding first Possibility ".Second, FEC academia are into final conclusion：" threshold effect is that the accuracy control system of all utilization error correcting codes is total to Have " incurable disease.W algorithms are thought：" how serious no matter disturb, ought to be better than not encoding using the accuracy control system of error correcting code System, be otherwise exactly to run counter to《Information theory》Macroscopic view it is cognitive ".Adhere to this conviction, W algorithms have found finally overcomes convolutional code " threshold effect " method, lower mask body describes in detail.
The present invention implements to provide a kind of forward error correction of convolutional code, with reference to Fig. 1, and methods described includes following step Suddenly：
S101, receive information source end send by the convolutional code sequence after noise jamming；The convolutional code sequence includes multiple Convolution word, the convolution word is that the information source end is generated according to generator matrix and it carries sequence of cells information and is superimposed upon Interference sequence in sequence of cells information.
Specifically, information source end is sent to stay of two nights end by information is transmitted, the meeting during the transmission information issues stay of two nights end There is distorted signals and noise jamming, therefore, information source end is first according to convolutional code production before the transmission information is sent It is (i.e. a kind of anti-dry that the information sequence (i.e. sequence of cells information) of the transmission information is converted to convolutional code sequence by (i.e. generator matrix) Disturb coding), the convolutional code sequence of conversion is then sent to stay of two nights end.
Stay of two nights end receives the convolutional code sequence, and convolutional code sequence now includes sequence of cells information, potentially includes noise Interference sequence.Under normal circumstances, convolutional code sequence is not only to have included sequence of cells information but also the interference sequence including superposition thereon, and Strong convergence characteristic is presented in above-mentioned convolutional code sequence.
As an alternative embodiment, setting a main decoder in the embodiment of the present invention, pass through the main decoder Perform above-mentioned steps 101.
S102, the sequence to be searched extracted in the convolutional code sequence.
It is dry from the separation of these convolutional code sequences in the threshold effect problem existed for existing encoder, the embodiment of the present invention Disturb before sequence, these convolutional code sequences are detected first, i.e., sequence to be searched is extracted from convolutional code sequence, and this waits to search Suo Xulie includes the bit of the first setting digit；As an alternative embodiment, above-mentioned first sets digit as n0It is whole Several times (the n of [3,1,8] convolutional code0=3).
As an alternative embodiment, setting an additional interpretations device in the embodiment of the present invention, translated by the auxiliary Code device performs above-mentioned steps 102.
S103, the smooth Penetrating air-flow for meeting according to the corresponding generator matrix of convolutional code directed euler graph rule, detect institute State whether sequence to be searched can accurately decode.
Specifically, each convolutional code sequence includes multiple convolution words, convolution word described in adjacent any two represent from The state of a upper time slot is transferred to the state of next time slot；Above-mentioned smooth Penetrating air-flow represent to have between convolutional code continuity and The continuity meets the constraint rule of generator matrix setting；
In specific detection method, if above-mentioned convolutional code sequence meets the smooth Penetrating air-flow of directed euler graph rule, Determine that sequence to be searched can be decoded accurately.
As an alternative embodiment, setting an additional interpretations device in the embodiment of the present invention, translated by the auxiliary Code device performs above-mentioned steps 103.
S104, if so, entering row decoding processing to the sequence to be searched, handling obtained coding sequence using decoding generates Obtained normal solution convolutional code sequence, and it is described by after noise jamming using the corresponding normal solution convolutional code covering of normal solution convolutional code sequence The corresponding convolutional code of convolutional code sequence.
Specifically, when above-mentioned convolutional code sequence meets the smooth Penetrating air-flow of directed euler graph rule, determining to be searched Sequence can be decoded accurately, now, and row decoding processing is entered to above-mentioned sequence to be searched, coding sequence is obtained, is then translated to this Code sequence carries out coded treatment, obtains normal solution convolutional code sequence, finally utilizes the corresponding positive deconvolution of normal solution convolutional code sequence Code covering is by the corresponding convolutional code of convolutional code sequence after noise jamming.
As an alternative embodiment, setting an additional interpretations device in the embodiment of the present invention, translated by the auxiliary Code device performs above-mentioned steps 104.Specifically, additional interpretations device can be decoded accurately in sequence to be searched, it can directly be determined Main decoder is invalid to the processing procedure of the sequence to be searched, and the normal solution convolutional code that later stage additional interpretations device is obtained directly covers master The result of decoder, and determine that the later stage enters row decoding processing to the normal solution convolutional code, determine the decoding of the normal solution convolutional code Output result is correct result.
S105, the strong convergence characteristic according to convolutional code sequence and the corresponding generator matrix of the convolutional code meet oriented Europe The smooth Penetrating air-flow of figure rule is drawn, subsequent interference sequence is separated in the convolutional code after being capped.
Specifically, multistage smooth type separates the convolutional code sequence that stay of two nights end is received, pollution and errorless convolutional code are found Region, wherein, errorless convolutional code region includes two kinds：First, convolutional code region is not completely by noise pollution；Second, volume Product code region is in itself by noise pollution but after polluting without needing the error correction still can be with correct decoding.
Therefore, in the embodiment of the present invention, multiple multistage smooth types for separating treatment convolutional code sequence can be built first, According to this multiple multistage smooth type, the interference sequence and the sequence of cells information are separated from convolutional code sequence.
In the embodiment of the present invention, in the new sequence to be searched of extraction and to the sequence to be searched in step 102 and step 103 The process detected is arranged, with the previous sequence to be searched by detection of sequence to be searched new to this in step 105 Processing procedure can be carried out simultaneously, improve the processing speed of decoding process.
As an alternative embodiment, setting a main decoder in the embodiment of the present invention, pass through the main decoder Perform above-mentioned steps 105.
S106, the interference sequence is entered row decoding processing, obtain maximum-likelihood decoding sequence.
Specifically, the mode matched according to the generator matrix with information source end enters row decoding processing to the interference sequence, obtain Maximum-likelihood decoding sequence.
As an alternative embodiment, setting a main decoder in the embodiment of the present invention, pass through the main decoder Perform above-mentioned steps 106.
The process for illustratively corresponding to step 101- steps 106 in the embodiment of the present invention is arranged below：
Point three parts are illustrated in the embodiment of the present invention, Part I：Different views in terms of cognition；Part II：Place Different ways in terms of reason；Part III：Decoder general diagram is summarized.
In decades, various FEC algorithms are thought：" what information source produced is some chance events, channel transmission be also with Machine event, interference is also certain chance event present in channel, it is therefore desirable to probability theory, random process so some count Mathematical tool is described and analyzed ".Therefore, current various decoding algorithms are all that probability based on " maximum likelihood rule " is translated Code algorithm.W algorithms are queried " therefore needing to be described and analyzed with probability theory, random process so some statistical mathematics instruments " Viewpoint.
W algorithms are thought：In any once specific communication process, the full detail that stay of two nights end is received all has occurred Determine event.No matter signal distortion and noise pollution how serious have, what stay of two nights end was received must be the data of one group of determination； And data that are still nonevent absolutely not, changing at random.For the determination event occurred, probability theory, random mistake should not be used Journey so some statistical mathematics instruments are described and analyzed.
Based on the cognition, the forward error correction of convolutional code provided in an embodiment of the present invention be based on mathematical algorithm, and Maximum likelihood folding coding algorithm based on probability theory in non-existing technology, can be right by method provided in an embodiment of the present invention The coding process of constraint length length is easy and effective and can solve the problem that the threshold effect problem that existing decoder is present.
Part I：Different views in terms of cognition
Second point：The optimal application occasion of probability theory is that still nonevent event is made prediction.Prediction effect is seen, generally Rate opinion can make rational judgement to high-probability event, and then helpless for low probability event, or even do what is made mistake Judge.Another application occasion of probability theory is to set up macroscopic theory, famous《Information theory》It is exactly that analysis is used as using probability theory The communication theory that instrument is set up, its development to information technology is generated persistently and deep effect.However, in specific information In processing event, often it is not belonging to the event of prediction or sets up macroscopic theory, probability analysis is not used as far as possible.This newest sight Point is not only applicable to FEC, it is likely that be also applied for other information process field.
W algorithms are thought：" in once specific communication process, the full detail that stay of two nights end is received all is to have occurred really Determine event ".Accordingly, W algorithms are further clear and definite：Convolutional code forward error correction technique is not belonging to handle chance phenomenon, and belongs to processing The problem of decisive phenomenon.Decisive phenomenon (phenomenon under certain condition, necessarily occurred) is the opposite of chance phenomenon；Deposit On the basis of mathematically Accurate Analysis, there is a possibility that accurately to decode.Thus inference：Using " probability theory, random process So some statistical mathematics instruments are described and analyzed " the decisive phenomenon of processing is to cause forward error correction technique complicated, decoding essence The low basic reason of degree.
Further, with reference to Fig. 2, the forward error correction of convolutional code provided in an embodiment of the present invention, in step 103, root Meet the smooth Penetrating air-flow of directed euler graph rule according to the corresponding generator matrix of convolutional code, whether detect the sequence to be searched Can accurately it decode, including：
S1031, the sequence to be searched is divided into multiple deconvolution time slot groups, meet multiple deconvolution time slot groups with Lower first imposes a condition；The sequence to be searched includes the bit of the first setting digit；Described first impose a condition for：Any two The individual adjacent deconvolution time slot group includes the overlapping bit of the second setting digit；
S1032, respectively multiple deconvolution time slot groups of division are carried out with deconvolution calculating using deconvolution formula, generation with The corresponding multiple coding sequences of the deconvolution time slot group；Each described coding sequence includes multiple warp cells；
Whether multiple warp cells overlapping in arbitrary neighborhood coding sequence are equal in S1033, the multiple coding sequences of detection；
S1034, if so, determining that the sequence to be searched can be decoded accurately.
Integrally above-mentioned steps 103 are illustrated with reference to step 1031~step 1034：In decades, various FEC Algorithm is described and analyzed using probability theory, random process so some statistical mathematics instruments.If being treated with mathematical angle FEC, it is found that most variable can obtain accurate solution.Only small part variable is difficult to directly extrapolate accurate solution, this kind of Variable as dependent variable again and other variable existence function relations (embodiment of the present invention is referred to as transfer function).Because convolutional code is The Memorability information source generated by sequential network, these dependent variables can be deduced according to transfer function.
Main decoder undertakes FEC whole error correction, work decoding.The information content once communicated is up to several megabits, and master translates Code device can produce decoding error unavoidably, based on the problem, increase additional interpretations device in the embodiment of the present invention, the additional interpretations device is Prevent and eliminate the key measure (therefore first illustrating additional interpretations device) that main decoder steps into erroneous path.
Additional interpretations device function introduction：
It is a mortal malady for decoding error correction procedure to step into erroneous path.Erroneous path be one it is progressive spread process, it is wrong Necessarily there is an initial point in path by mistake, and " checking erroneous ideas at the outset " is a kind of effective measures.Harsh is：Even if main design of encoder is again It is rigorous, still there is the danger for stepping into erroneous path, the effect of additional interpretations device is exactly to reduce this dangerous possibility.
Additional interpretations device includes two functions：
First function：13 coding sequences are constantly generated using the deconvolution formula of 13 bits, and are continuously compared 13 coding sequences (referring to table 1) of front and rear generation.
Table 1：The function description of additional interpretations device
B (3X-24)~b (3X+26) in table 1 is contaminated area, but b therein (3X+0)~b (3X+26) 27 bits claim For " potential pollution-free area " (possible errorless dye area).
Assuming that the b (3X+0) of moment~b (3X+26) 27 bits are not contaminated, 1~5 time slot in table 1 will be anti-respectively The respective 13 decoding series of convolution generation, and 12 cells of adjacent time-slots are overlapping by equal, when 5 time slots are fully met During this condition, the normal solution of S (X+0)~S (X+16) 17 sequence of cells information is comparatively reliable obtained.If be unsatisfactory for This judges that b (3X+0)~b (3X+26) 27 bits are necessarily contaminated.This is that have to connect according to the point on Euler diagram and side What coherence was determined；Wherein, the point on above-mentioned Euler diagram and the point in corresponding convolutional code and while.
Second function：When fully meet 12 cells of adjacent five time slots by it is equal when, be doomed potential errorless dye Area b (3X+0)~b (3X+26) is not contaminated really, and 17 coding sequence S (the X+0)~S (X+16) for the warp generation that is bound to is Whole normal solutions.According to the coding sequence of 17 normal solutions, errorless convolutional code sequence B (3X-24)~B (3X+ of 51 bits can be generated 26)；The errorless convolutional code sequence B (3X-24) of 51 bits~B (3X+26) is obtained using errorless 27 bit convolutions code, is filled Divide the technological value for embodying coding gain.Additional interpretations device is forced to cover the line of main decoder with B (3X-24)~B (3X+26) Road convolutional code b (3X-24)~b (3X+26), it is ensured that main decoder is returned on correct decoding path.Additional interpretations device is gram Take the effective measures of main decoder " threshold effect ".
In addition to 13 bit deconvolution formula, W algorithms also provide other two groups of deconvolution formula, three groups of court verdicts Repeatability is crucial.With the continuous deterioration of interference environment, continuously >=27 bit interference-free coverage area will be reduced therewith.Valuable is " deconvolution is equivalent " phenomenon that the second point of two parts is proposed compensate for this problem, and it is " anti-that objective reality part >=27 bits Convolution is equivalent " interference pattern, they, which can equally be played, prevents " threshold effect " from acting on.
In Fig. 8, the half of the additional interpretations device lower right corner adds the comparison of matrix three groups of warp sequence of cells information of completion.Usual feelings Under condition, the deconvolution result of three groups of formula is different, and half adds matrix to close, and " covering control B ", which is enabled, " covers control A ", by main decoding Device implements covering function.When three groups of deconvolution form identical >=17 cells, convolutional code sequence is necessarily received Occur in that the noiseless region of >=27 bits.Half adds matrix to close, and " covering control A ", which is enabled, " covers control B ", 17 letters of this group The errorless convolutional code of 51 bits of member regeneration covers corresponding circuit convolutional code section, forces decoder to return to correct decoding road On footpath, FEC " threshold effect " efficiently avoid.Additional interpretations device can quickly, the function of Exact recovery correct path, protect Having hindered main decoder persistently will not rest on erroneous path.
At present, academia thinks that " convolutional code is that the correlation leaned between adjacent-symbol provides protection, convolutional code antiburst error Poor ability "；In this regard, W algorithms have different understanding and practice result.W algorithms are thought：To verify the anti-interference energy of convolutional code Power, must be by the way of " interference complete or collected works ", and so-called " interference complete or collected works " are exactly the interference pattern of traversal.The embodiment of the present invention will be used " deconvolution is equivalent " and " error correction is equivalent " 2 points of proof convolutional code antiburst errors are very capable, very poor absolutely not.
Before proof, [3,1,8] convolutional code partial content related to directed euler graph is quoted from first：
The path for going over each summit is exactly Euler Access (loop), and the figure that there is Euler's circuit is exactly Euler diagram, and [3, 1,8] convolutional code complies fully with the feature of above-mentioned Euler diagram.
Specifically, the state value of sequence of cells information is considered as the summit of Euler diagram, [3,1,8] convolutional code be then relative to Euler Access on Euler diagram.[3,1,8] convolutional code has 256 summits, including two special summit A and B.Special summit A is 8 displacement storage [0,000 0000] states, and special summit B is [1,111 1111] state.In practice, generally spy Different summit A is defined as starting point (starting point is the original state or holding state of communicating pair).
W algorithms are proved：The relevance of decoding algorithm and signal to noise ratio is smaller, and decoding effect is better (table 2).For that purpose it is necessary to The original state for making full use of stay of two nights end errorless, reduces the factor for being related to contaminated area in deconvolution formula as far as possible.Table 2 is listed Utilize four public affairs of [S0, S2, S4, the S12] of 7 bit deconvolution, 13 sequence of cells information of [b27~b33] of contaminated area Formula.
Though W algorithms do not accept " asymptotic coding gain value ", the concept of coding gain is to recognize that.Due to realizing 7 bits 13 sequence of cells information of deconvolution, coding gain=10log (39/7)=7.46dB significantly breaches the theory decoded firmly Asymptotic value 4.77dB, has shaken the existence of " theoretical asymptotic value γ (dB) " at all.
Illustrate 3：24 bit full 0s of stay of two nights end original state are a good beginning, if constantly obtaining contaminated area A series of cell normal solution, it is possible to continue backward always without mistaken ideas initial, so as to continue with 7 bit deconvolution The technical advantage of 13 sequence of cells information.It must declare：" constantly obtaining a series of cell normal solution in contaminated area " is all The special of the highest goal that FEC is unanimously pursued, absolutely not W algorithms is made excessive demands.
Illustrate 4：In addition to foregoing 13 bit warp, 13 cells, related bits≤7 of the deconvolution formula of table 2 are compared Spy, is a kind of higher efficiency " the equivalent phenomenon of deconvolution "." the equivalent phenomenon of deconvolution " of table 2 refers to entirely different 256 The decoding result of the deconvolution generation of interference pattern is identical！Highly versatile generally is represented with " drawing inferences about other cases from one instance ", is then had this moment " the effect of act one anti-256 ".It not only improves decoding efficiency, while considerably reducing the workload of quantitative analysis.
Deconvolution is equivalent：There are 256 with the fully equivalent interference pattern of glitch-free 12 bit [full √] pattern, table 3 is arranged 6 therein are lifted.Maximum interference strength is 83.3% in table 3, so strong interference unexpectedly with glitch-free [full √] figure Case deconvolution is equivalent, and ([S8~S12] five cells of deconvolution generation are identical！)；Further with glitch-free 18 bit The fully equivalent interference pattern of [full √] pattern has 4096." deconvolution is equivalent ", which embodies convolutional code, has extremely strong convergence Property.The viewpoint that it has shaken at all " coding gain ... with signal to noise ratio about "；Especially shaken " convolutional code be lean on adjacent-symbol between Correlation provide protection, convolutional code antiburst error poor ability " viewpoint.The phenomenon of " deconvolution is equivalent " is that W algorithms are true The decoding rule of " error correction after first decoding " is found.
2. error correction is equivalent：The equivalent mapping relations being conceived between interference and decoding cell of the deconvolution of table 3.Table 4 entangles It is wrong equivalent, it is to focus in convolutional code sequence whether disturbing, judges whether each in code stream is disturbed.
The error correction equivalent scope that W algorithms can be realized is than larger, and operating range is across more than 60 bits (can also realizing more On a large scale).The small range of table of the embodiment of the present invention 4 judges to be introduced.
Illustrate 5：Table 4 only shows continuous item, eliminates outlier.For example judge the b28 of contaminated area I, be only decided by pollution The b34b35 in area II, as long as b34b35 is not interfered or disturbed by even number, can accurately judge b28 whether by To interference.
Illustrate 6：B9~b23 of original state belongs to initial without mistaken ideas, if the initial related bits deviation quilt without mistaken ideas Calculate as 1 (referred to as lodging a false accusation against phenomenon), necessarily the related bits of contaminated area I, II have the interference of odd number, and this is one complementary Error correction determination methods.
Thirdly：To verify the antijamming capability of convolutional code, W algorithms are by the way of traversal " interference complete or collected works ".Therefore, The ability for asking W algorithms that there is accurate identification interference pattern.
Illustrate 8：Table 5 lists 7 formula, can be the 2 of [b27~b38]12=4096 interference patterns are exactly divided into 27 =128 general character subsets.Accurate to judge 128 general character subsets, each general character subset correspond to 16 specific 12 bit interference Pattern.It can be seen that W algorithms are substantially better than probability decoding algorithm.(step locks 16 suspicion interference figures to the mono- Ming Liao of first, method Jian Case, wherein necessarily including actual interference pattern).Second, judged result and interference strength it is substantially unrelated (W algorithm foundations it is orthogonal Judge, rather than the bit error rate).
Illustrate 9：Please note that [b24~b26] of table 5 is denoted as ◆, they are the beginnings (arch-criminal) of contaminated area, and W is calculated Method is referred to as " byte of beheading ", and inerrably the correct byte of beheading that handles is the guarantee and shortcut constantly accurately decoded.But " cut First byte " is difficult to directly accurately judge, it is necessary to using a variety of orthogonal comprehensive descisions.
4th point：Using the regularity and relevance of anti-interference coding, solution most preferably recovers raw information.
W algorithms are thought：The FEC technologies of convolutional code belong to decisive phenomenon, and decisive phenomenon answers the accurate solution of existence anduniquess. It is cell S8 that " byte of beheading " is corresponding in table 5, and contaminated area [b27~b38] has 212=4096 interference patterns, W algorithms are borrowed Help the deconvolution formula to calculating S8 of 7 bits.Specific method is 2 [b27~b38]12=4096 interference patterns are divided into two Subset (subset A, B), subset A and subset B include 2048 interference patterns, corresponding subset A 2048 interference patterns, 7 respectively The deconvolution formula Accurate Estimation of bit goes out S8 normal solution.Corresponding subset B 2048 interference patterns, the deconvolution of 7 bits is public Formula Accurate Estimation goes out S8 anti-solution.More importantly：In addition to S8 judgement is different, subset A and subset B's amounts to 256 S9~S12 of equivalent interference pattern deconvolution generation is identical！Correctly judge that interference pattern belongs to the shortcut that subset A or B are error correction, But interference pattern belongs to subset A or B and is difficult to accurately judge (there is possibility).
It is obvious that W algorithms have avoided the error correction thinking being difficult to, because the quantity of interference pattern is very big, accurately sentence Break each interference pattern, and be modified and be nearly impossible thing.Only classify by its general character, directly set up decoding letter Mapping relations between member and interference pattern, can significantly simplify the complexity of FEC decodings, improve decoding precision.
Illustrate 10：16 interference legends in subset A are listed in table 6, from 12 bits noiseless (full √) to 12 bits Middle to there are 11 interference, interference strength rises to 0.92% from 0.If according to the error correction thinking of probabilistic decoding, it is difficult to which how imagination is sentenced Break and correct so numerous, complicated, strong interference！Most preferably recovery raw information how is realized againSo, at first In the 7th point divided, W algorithms are thought：The appellation of forward error correction (FEC) deviate from final goal, because the thinking of error correction is difficult to reality Existing (or even can not realize), therefore be unfavorable for recovering raw information.Table 6 fully illustrates：Processing procedure ≠ final goal, due to It is fuzzy in understanding, form the notable difference of interpretation method and effect.
Thirdly：In FEC fields, the describing mode of widely used code tree figure or generator matrix.For different anti-interference Coding, code tree figure or generator matrix are described as a tree structure either semi-infinite matrix for diverging without exception, it is clear that can not It is convincing.Viterbi professor is using grid chart (also referred to as hedge figure) description convolutional code, and the convolutional code described by it presents receipts The feature held back.But, the concept that convolutional code is described, analyzed with grid chart is extremely complex；The net of simplest (2,1,2) convolutional code Trrellis diagram using four row multiple rows description；Understand on numerous and diverse grid chart that each node is extremely difficult with path change, less use Say discovery and understand error correction method, in this regard, with reference to Fig. 3, forward error correction provided in an embodiment of the present invention, step 104, root Meet the smooth of directed euler graph rule according to the strong convergence characteristic and the corresponding generator matrix of the convolutional code of convolutional code sequence Subsequent interference sequence is separated in Penetrating air-flow, the convolutional code after being capped, including：
S1041, according to be used for generate convolutional code generator matrix maximum freely away from, build for described in separating treatment roll up Multiple multistage smooth types of product code sequence；The smooth type includes：Unrelated bit and for convolutional code sequence described in separating treatment Related bits；The related bits include the characteristic parameter for being used to detect the convolutional code sequence：It is each described multistage smooth The related bits difference that formula includes, and the characteristic parameter of different related bits detections is different；The correlation Bit is the point in any a line that the Euler diagram includes；The unrelated bit then for the Euler diagram include it is any one Point outside bar side.
S1042, according to multiple multistage smooth types, the subsequent interference sequence is separated from the convolutional code sequence and described Sequence of cells information.
Overall description is carried out to above-mentioned steps 104 with reference to 1041~step 1042 of above-mentioned steps：C.E. celestial beings in 1948 Agriculture exists《A mathematical theory of communication》In, illustrate the method that reliable communication is realized in channel is thanksed for your hospitality, it is proposed that famous thankss for your hospitality Channel coding theorem, has established the foundation stone of anti-interference coding.Hereafter, due to the attention of communication work person and mathematician, in theory A variety of anti-interference codings and its efficient coding technology are generated in practice.
FEC belongs to high reliability communication issue, is to study the regularity and relevance that how to utilize anti-interference coding, solves Most preferably recover raw information, realize the technology of reliable communication.It must be emphasized that：Error correction be recover raw information important means and Non-final purpose.W algorithms are thought：To a certain extent, the appellation of forward error correction (FEC), which deviate from, " utilizes anti-interference coding Regular and relevance, solves most preferably to recover raw information " final goal.Be conducive to recovering former because error correction is one The processing procedure of beginning information, rather than high reliability communication purpose.It should be emphasized that：In processing procedure ≠ final goal, understanding It is fuzzy, the huge difference (analysis for referring to Part II of the embodiment of the present invention) of interpretation method will be caused.
W algorithms are advised：In terms of the overall goal of high accuracy decoding, forward error correction (Forward Error Correction) May be more reasonable, more accurate to Information recovering (Forward Information Renovation) before being renamed as.
Part II：Different ways in terms of processing
First point：From the convolutional code sequence of reception, separation contaminated area with without mistaken ideas be FEC matter of utmost importance
Whole receiving sequence is scanned with smooth type, contaminated area is therefrom isolated and without mistaken ideas.An actual traffic can Several megabits can be reached, interference pattern is ever-changing.To realize tight monitoring effect, W algorithms employ 11 smooth types (prison Width is regarded still as 24 bits), separating effect is close to 100%.In fact, only 24 continuous (0000 ... 00000) are only pure Be not contaminated situation, such case only accounts for the 2 of whole interference pattern complete or collected works24One of power.
Whether identification, which receives data and be contaminated, is divided into two kinds of situations, the first, accurately judge to receive the pollution in data Area, this is the situation of the overwhelming majority.Second, it is contaminated without finding (embodiment of the present invention is referred to as failing to judge).It is general and Speech, the span of smooth type is longer, more, and possibility of failing to judge is lower.But, even if using longer, tighter smooth type, still In the presence of failing to judge, the embodiment of the present invention is referred to as in " pseudo- pollution-free area ".
For failing to judge, phenomenon be need not worry overmuch, as long as formula is tight enough, can be also found even if very small interference.Really What is failed to judge is the interference pattern with evident regularity that (table 7) is listed on the contrary.What is more important：Pseudo- pollution-free sequence and nothing There is general character phenomenon between [full √] sequence of interference.So-called general character phenomenon refers to：Pattern of failing to judge [1~4] has been superimposed with not having There is " deconvolution is equivalent " in one general character subset of reception convolutional code Sequence composition of superposition interference [full √], same general character subset Or the sequence of cells information or error correction pattern of " error correction is equivalent " phenomenon, the i.e. recovery that they judge are substantially the same.Maximum in table 7 Interference strength be 75%, so strong interference has general character with noiseless unexpectedly, has shaken that " coding gain ... has with signal to noise ratio The viewpoint of pass ".
W algorithms are thought：Just like language be interpersonal communication instrument it is the same, code tree figure or generator matrix describing mode divide absolutely not Analyse serviceable tools in terms of FEC；Maximum free distance dfBe conducive to the error correction of decoder, but to understand numerous and diverse grid The change in each node and path is extremely difficult on figure, it is more difficult to finds and understands error correction method.So, grid chart is not still reason The instrument thought.
4th point：W algorithms are thought：" generator matrix of convolutional code meets the feature of directed euler graph ".In directed euler graph On, it is desirable to " pen is not from paper " by each node and its link adjacent node between side.In information source end, convolutional code is not folded Plus any interference, sequence of cells information generation convolutional code process is exactly the process of one " pen is not from paper ", the convolutional code formed It is a series of data flow of mellow and full insertions.There is distortion and noise jamming in transmitting procedure, stay of two nights end is typically to be contaminated Convolutional code information sequence, lose the feature of mellow and full insertion, show the hopping sense of " pen leaves paper ".
W algorithms are thought：In same n0k0Under, the more important directions of construction convolutional code are to find " most short smooth type ".Because The span of most short smooth type is shorter, and each high-order smooth type is also more compact, beneficial to the various interference patterns of differentiation.W algorithms have been analyzed： According to the nonsystematic convolutional code that the maximum freely production away from construction is G (0)=[557], G (1)=[663], G (2)=[711] Most short smooth type be 12 bits.If reconfiguring production according to the concept of directed euler graph, it is possible to form shorter Smooth type.At present, W algorithms only continue to use the established maximum free distance d of industryf=18 convolutional code productions.
Multistage smooth type separates the convolutional code sequence that stay of two nights end is received, and finds pollution and errorless convolutional code region [table 8]。
Illustrate 1：B (3m+0)~b (3m+23) is convolutional code sequence number；Mark in table 8 " 1 " for related bits；Mark " 0 " For unrelated bit.In smooth type, related bits are the points on a series of convolutional code data trajectory mark of mellow and full insertions；It is unrelated Bit is then the point outside track.This is the concept that W algorithms are proposed first in FEC fields.Famous Viterbi decoding algorithm is Optimal probability decoding algorithm based on hedge figure (Trellis).Its decoding algorithm is exactly to calculate, compare on grid chart, tiring out Each paths are counted, the minimum path of an error code is therefrom selected.Viterbi decoding algorithm does not have the concept of related bits, all Convolutional code participates in ACS processing procedures without exception, increases the ambiguity of judgement.
Actual analysis shows：12 bit smooth types can be found that 50% Polluted area；If being put down for two using 12,15 bit Gliding style, it can be found that 75% interference；87.5% interference is then can be found that according to 12,15,18 bit, three smooth types, with The length increase of smooth type, the effect of filtering is also incremented by rapidly.5 smooth types in table 8 can be found that 96.9% pollution Region；Atomic small Polluted area is can be found that according to more smooth types.The total span of the discrimination formula of table 1 is 24 bits (embodiment of the present invention is referred to as monitoring width).
The generator matrix for proving convolutional code meets directed euler graph, it is therefore intended that more clearly analyze the feature of convolutional code, To find convolutional code, all are potential regular；And it is not limited to research Euler's closed problem (by each edge in figure once and only Once, and excessively each summit loop).It is a kind of quite preferable to be described and analyzed convolutional code by the way of directed euler graph Instrument, contribute to the error correction formula for finding, searching for, analyzing convolutional code.
Further, with reference to Fig. 4, in the embodiment of the present invention, in step 1041, according to the generation square for generating convolutional code The maximum of battle array freely away from, build for the multiple multistage smooth of convolutional code sequence described in separating treatment, including：
1041a, according to the maximum of the corresponding generator matrix of the convolutional code freely away from the most short multistage smooth type of structure.
5th point：In FEC theory narrative, sequence of cells information is often considered as one with the interference sequence being superimposed thereon Individual entirety is described and analyzed, and causes the complication of FEC problems.
W algorithms are thought：The convolutional code sequence generated by sequence of cells information is not one with the interference sequence being superimposed thereon It is overall.To a large extent, two independent parts can be divided into.Multistage smooth type is both ideal tools of separation, Both separation are apparent from FEC.Highlighted FEC error correction of interest and be only limitted to interference sequence part, and with the letter of information source Cease sequence unrelated.
The method of multistage smooth type agree without prior consultation in《And member opinion》.《And member opinion》" non-sine orthogonal transformation is handling discrete letter to title Number combination property far better than Fourier transform, and most non-sine orthogonal transformations all by and member opinion based on "." number In word communication system, noise is actually noise sequence and information sequence and first mistake being added to the interfering process of channel Journey ".
Using the strong convergence characteristic and Euler diagram feature of convolutional code sequence, the disturbed situation of received information sequence is found.
Using the strong convergence characteristic and Euler diagram feature of convolutional code sequence, the formula of interference is corrected in search.
Using simultaneously member opinion, the method for performing received information sequence " with wrong error correction ".
Using simultaneously member opinion, directly recover the method for received information sequence.
Last decade, " the number solely game " in the fashionable world is originating from Euler's invention《Latin square》, it is tightly to be patrolled The ideal tools of volume computing, the logical formula that W algorithms are deduced agree without prior consultation completely in《Latin square》.Usurped wantonly if interference is considered as Change " correctly counting only matrix ", its result is necessarily filled with flaws.
The n that the embodiment of the present invention is discussed0,k0, M=[3,1,8] convolution coder convolution coder constant when belonging to, Therefore the formula that the embodiment of the present invention is provided can on a large scale be verified in [3,1,8] nonsystematic convolutional code.
To illustrate clear accurate purpose, the formula and form for arranging the embodiment of the present invention are described as follows：
S (m) represents sequence of cells m cells of information；
B (3m+0)~b (3m+2) represents the 3 bit convolution code elements of correspondence S (m) cell；
W (m) represents the convolution word of correspondence S (m) cell, i.e. (3m+0)~(3m+2) 3 bit symbols；
The embodiment of the present invention is generally described using binary system, is then described for longer data using octal system.According to number According to the rule of communication, information sequence first sends low level, and the counting of 8 systems is also from low level to high-order evaluation.By this agreement：S (10) W10 b30, b31, b32 is correspond to, if W10=2 represents that b30b31b32=010, W10=4 represent b30b31b32= 001。
1041b, on the basis of the most short multistage smooth type, builds multiple multistage smooth types；Each is described multistage flat The exponent number of gliding style is all higher than the exponent number of the most short multistage smooth type.
Four following smooth types are now provided to verify for reader.
S (m)=b (3m-2) ⊕ b (3m-1) ⊕ b (3m+2) ⊕ b (3m+4) ⊕ b (3m+5) formula 1
Formula 1 is the deconvolution formula for directly recovering cell S (m) with convolutional code.The formula includes 5 related bits, only Will 5 related bits be not interfered or disturbed by even number, can obtain S (m) normal solution.
Smooth type:[30425] formula 2
Formula 2 is the smooth type of 15 bits.
Smooth type:[614000023] formula 3
Formula 3 is the smooth type of 27 bits of 12 unrelated bits of a leap.
Smooth type:[51324000000556] formula 4
Formula 4 is the smooth type of 51 bits of 18 unrelated bits of a leap.
2~formula of formula 4 is used to find whether convolutional code sequence is disturbed, and does not interfere with or even number interference occurs When, smooth type result is 0, is otherwise 1.Sufficient smooth type can be found that extremely micro interference, realizes mask data cell The free-revving engine of sequence information and interference sequence, advantage is provided for forward error correction.
W algorithms can provide all kinds of smooth types of more than hundreds of, the application value of which part formula it is very high (referring to The content of additional interpretations device).In view of W algorithms are the method decoding of pure mathematics, it is described naturally by the way of formula.In reality In the decoder of border, these formula are fully converted to specific logic circuit.In summary, " generator matrix of convolutional code, which meets, to be had To Euler diagram " the great practical value of viewpoint, it is the optimized analysis instrument in FEC fields, is also the decoding side of stay of two nights end most preferably Method.
Further, the forward error correction of convolutional code provided in an embodiment of the present invention, the convolutional code be information source end by Generated according to the generator matrix of [3,1,8] nonsystematic convolutional code；
The most short multistage smooth type is [3375], and [3375] are quadravalence smooth type and corresponding related bits and unrelated ratio Spy is followed successively by：110、110、111、101；The multistage smooth type at least includes：Nine rank smooth types [614000023] and ten quadravalences Smooth type [51324000000556].
6th point：At present, academia thinks：" coding gain is the important indicator for evaluating error correction code performance ".And define volume Code gain be：" under a certain bit error rate, after certain specific error-correcting code system, letter is can be obtained when being relatively free of application Make an uproar than reduction decibels ".While two conclusions are proposed, and first, " coding gain is not only relevant with error correcting code, and and noise Than relevant ".Second, " when signal to noise ratio is very low, coding gain is changed into negative value, and (ratio is not using the performance after error correcting code for this explanation Coding) it is worse.This threshold effect is common to the accuracy control system of all utilization error correcting codes ".This conclusion makes all FEC error correcting codes are overshadowed, are the inhibitions on " accuracy control systems of all utilization error correcting codes " head！W algorithms approval " coding Gain is the important indicator for evaluating error correction code performance " viewpoint, but W algorithms query two subsequent conclusions, to two conclusions It is analyzed as follows：
If " coding gain is relevant with signal to noise ratio ", following antinomy certainly will be produced；" when signal to noise ratio is very low, coding increases Benefit is changed into negative value ".When signal to noise ratio very high (even without interference), coding gain is also negative value.Because when not interfering with, no The system of coding does not have error code, and the system encoded wastes bandwidth.Comprehensive conclusion is：Signal to noise ratio is very low or very high Two kinds in the case of, coding gain is all changed into negative value.People can't help asking：Coding gain reaches most preferably in what situationsIt is theoretical Boundary does not provide clear and definite answer.
As for " threshold effect ", clear and definite solution (refers to the embodiment of the present invention the 3rd to the embodiment of the present invention Point).It is exactly the inhibition for relieving the accuracy control system using convolutional code error correction to overcome threshold effect, makes us trusting courageously Ground utilizes and improved the accuracy control system of convolutional code error correction.The embodiment of the present invention wishes that the thinking of W algorithms is compiled to other error correction Code system is helpful.
7th point：Academia is for specific error correcting code, it is proposed that theoretical asymptotic coding gain value γ (dB).Asymptotic value It is the desired result under specified conditions, typically represents the ideal value of " infinitely approach and be but unreachable to ".
W algorithms query the viewpoint of " asymptotic coding gain value γ (dB) ".W algorithms have been accomplished：According to 13 errorless bits Convolutional code sequence can accurately translate coding gain=10log (39/13) of 13 sequence of cells information, i.e. convolutional code= 4.77dB, has reached the theoretical asymptotic value of hard decoding, has shaken the existence of theoretical asymptotic value γ (dB).
W algorithms are further proved：The relevance of decoding algorithm and signal to noise ratio is smaller, and decoding effect is better.In table 9, letter The related deconvolution bit of first S (m+2) only has b (3m+4), b (3m+5), b (3m+8), b (3m+10), 5 bits of b (3m+11), Other 8 are unrelated bits.Show that the relevance of decoding algorithm and signal to noise ratio is smaller, decoding effect is better.Need to again point out： If " 5 bits related to cell S (m+2) are not all interfered or disturb and (bear as just) by even number, instead The cell S (m+2) of convolution generation is accurate decoding result ".
Illustrate 2：B (3m+1)~b (3m+13) is convolutional code sequence number；S (m-8)~S (m+4) is to utilize 13 bit convolutions code 13 cells of deconvolution generation.Mark in table 9 " 1 " for related bits；Mark " 0 " for unrelated bit.In deconvolution formula In, related bits are the convolution code bits closely related with decoding result；Unrelated bit be then and decoding the incoherent convolution of result Code bit.In short, unrelated bit is regardless of whether unrelated with decoding result by interference.
In actual FEC processing procedures, in the convolution code stream at stay of two nights end, there is the chance in the errorless region of 13 bits still Compare high, using table 9 formula can a deconvolution produce 13 correct sequence of cells information, equivalent to spanning The interference effect of b (3m-23)~b (3m+0) 24 bits before (have ignored in other words), decoding efficiency is considerable.
Further, with reference to Fig. 5, in the forward error correction of the convolutional code provided in the embodiment of the present invention, step 105, The flat of directed euler graph rule is met according to the strong convergence characteristic of convolutional code sequence and the corresponding generator matrix of the convolutional code In sliding Penetrating air-flow, the convolutional code after being capped before separation subsequent interference sequence, in addition to：
S1051, the sequence to be fractionated extracted in the convolutional code sequence；The sequence to be fractionated includes the 3rd setting digit Bit；
S1052, classification processing carried out to the sequence to be fractionated according to the characteristic parameter of convolutional code sequence, obtain multiple many Level sequence；Each described multistage sequence corresponds to the interference pattern of the first setting number；The interference pattern character pair ginseng Number has general character；The characteristic parameter at least includes：Warp equivalence value and error correction equivalence value；The warp equivalence value is to described The sequence of cells information that sequence progress warp processing to be fractionated is obtained is identical.
Further, with reference to Fig. 6, the forward error correction of convolutional code provided in an embodiment of the present invention, step 106, to institute State interference sequence and enter row decoding processing, after obtaining maximum-likelihood decoding sequence, in addition to：
S1061, the maximum-likelihood decoding sequence progress convolutional encoding processing to obtaining, obtain covering convolutional code sequence Row.
S1062, using it is described covering convolutional code sequential covering by the convolutional code sequence after noise jamming.
Due to there is the powerful reliable support of additional interpretations device, main decoder does not almost have to concern that decoder steps into mistake The danger in path by mistake.But, with the continuous deterioration of interference environment, continuously >=27 bit interference-free coverage area will be reduced therewith, be aided in The function of decoder will slightly have to a certain extent to be reduced.Main decoder must make full use of the regularity of the convolutional code of discovery, The scope of Exact Solutions is compressed using multiple orthogonal formula, increase obtains the possibility accurately solved, actively avoids stepping into erroneous path.
Determine that convolutional code forward error correction technique is not belonging to handle chance phenomenon in the embodiment of the present invention, and belong to processing and determine The problem of property phenomenon.The decoding precision of W algorithms is proportional to " monitoring width ", " monitoring width " more big available orthogonal formula and rule Rule property is more, and decoding precision is also higher.
It is quite big to obtain workload under accurate solution, the platform of pure mathematical method.Tracing it to its cause has three.First, current Convolutional code production is produced according to " freely maximum away from ", may not meet the regularity of " most short smooth type ".Second, W algorithm Decoding precision be proportional to the scale of interference complete or collected works, the more big available orthogonal formula of scale and regular more, the only matrixes of number It is more big tighter.3rd, convolutional code generator matrix has symmetry, thus the limit range of Exact Solutions is two.
This is limited in, from the point of view of versatility, W algorithms there will be three classes to calculate result, and the first kind is accurate judgement；Equations of The Second Kind is It is accurate to judge (more than two finite solutions)；3rd class is made a choice judgement according to the interference strengths of Exact Solutions.
The forward error correction of convolutional code provided in an embodiment of the present invention, is extracted in convolutional code sequence using the first mechanism Sequence to be searched, and meet according to the corresponding generator matrix of convolutional code the smooth Penetrating air-flow of directed euler graph rule, detection is treated Whether search sequence can accurately decode；If so, treating search sequence enters row decoding processing, covered using obtained normal solution convolutional code Lid is by the corresponding convolutional code of convolutional code sequence after noise jamming, then using strong convergence of second mechanism based on convolutional code sequence Characteristic and the corresponding generator matrix of convolutional code meet the smooth Penetrating air-flow of directed euler graph rule, the convolution after being capped Subsequent interference sequence is separated in code, and enters row decoding processing to the subsequent interference sequence, maximum-likelihood decoding sequence is obtained, with showing There is the decoder based on maximum likelihood folding coding algorithm in technology, improve the means of decoding performance for constraint length length The problem of coding process is complicated and decoder has threshold effect in itself is compared, and it is based on mathematical thought, using the second machine Make the transmission convolutional code to information source end in the way of error correction after first decoding and enter row decoding processing, decoding speed is fast and the degree of accuracy is high； And also start the first mechanism in real time to be monitored the first mechanism, supervise the second mechanism to maintain on correct decoding path, The problem of the second mechanism steps into erroneous path is effectively weakened, the threshold effect for efficiently solving only one mechanism of startup is asked Topic.
The embodiments of the invention provide a kind of fec arrangement of convolutional code, described device is used to perform a kind of convolutional code Forward error correction, with reference to Fig. 7, described device includes：
Receiving module 11, for receive information source end transmission by the convolutional code sequence after noise jamming；The convolutional code sequence Row include multiple convolution words, and the convolution word is that the information source end is generated according to generator matrix and it carries sequence of cells letter Breath and interference sequence；
First abstraction module 12, for extracting the sequence to be searched in the convolutional code sequence；
Detection module 13, the smooth Penetrating air-flow of directed euler graph rule is met according to the corresponding generator matrix of convolutional code, Detect whether the sequence to be searched can accurately decode；The smooth Penetrating air-flow has continuity and institute for the convolutional code State the constraint rule that continuity meets generator matrix setting；
First overlay module 14, for when the sequence to be searched can be decoded accurately, entering to the sequence to be searched Row decoding processing, the normal solution convolutional code sequence that obtained coding sequence generation is obtained is handled using decoding, and utilize positive deconvolution The corresponding normal solution convolutional code of code sequence covers the corresponding convolutional code of convolutional code sequence by after noise jamming；
Separation module 15, for the strong convergence characteristic according to convolutional code sequence and the corresponding generator matrix of the convolutional code Meet the smooth Penetrating air-flow of directed euler graph rule, subsequent interference sequence is separated in the convolutional code after being capped；Institute Stating convolutional code sequence includes multiple convolution words, and it is next that the adjacent convolution word of any two represents that the state of a time slot is transferred to The state of time slot；The smooth Penetrating air-flow represents that convolutional code has continuity and the continuity meets generator matrix setting Constrain rule；
Processing module 16 is decoded, for entering row decoding processing to the subsequent interference sequence, maximum-likelihood decoding sequence is obtained Row.
Further, the fec arrangement of convolutional code provided in an embodiment of the present invention, detection module 13, including：
Division unit, for the sequence to be searched to be divided into multiple deconvolution time slot groups, makes multiple deconvolution time slots Group meets following first and imposed a condition；The sequence to be searched includes the bit of the first setting digit；Described first imposes a condition For：The adjacent deconvolution time slot group of any two includes the overlapping bit of the second setting digit；
Generation unit, based on carrying out deconvolution to multiple deconvolution time slot groups of division respectively using deconvolution formula Calculate, generate multiple coding sequences corresponding with the deconvolution time slot group；Each described coding sequence includes multiple warps Cell；
Detection unit, for detecting that multiple warp cells overlapping in arbitrary neighborhood coding sequence in multiple coding sequences are It is no equal；
Determining unit, for multiple warps letter overlapping in arbitrary neighborhood coding sequence in detecting multiple coding sequences When member is equal, determine that the sequence to be searched can be decoded accurately.
Further, the fec arrangement of convolutional code provided in an embodiment of the present invention, separation module 15, including：
Construction unit, for according to the maximum for the generator matrix for being used to generate convolutional code freely away from building at separation Manage multiple multistage smooth types of the convolutional code sequence；The smooth type includes：Unrelated bit and for described in separating treatment roll up The related bits of product code sequence；The related bits include the characteristic parameter for being used to detect the convolutional code sequence：It is each described The related bits difference that multistage smooth type includes, and the characteristic parameter of different related bits detections is different； The related bits are the point in any a line that the Euler diagram includes；The unrelated bit then includes for the Euler diagram Any a line outside point；
Separative element, for according to multiple multistage smooth types, separated from the convolutional code sequence interference sequence and The sequence of cells information.
Further, the fec arrangement of convolutional code provided in an embodiment of the present invention, construction unit, including：
First builds subelement, free away from structure is most short for the maximum according to the corresponding generator matrix of the convolutional code Multistage smooth type；
Second builds subelement, on the basis of the most short multistage smooth type, building multiple multistage smooth types；It is each The exponent number of the individual multistage smooth type is all higher than the exponent number of the most short multistage smooth type.
Further, the fec arrangement of convolutional code provided in an embodiment of the present invention, the convolutional code be information source end by Generated according to the generator matrix of [3,1,8] nonsystematic convolutional code；
The most short multistage smooth type is [3375], and [3375] are quadravalence smooth type and corresponding related bits and unrelated ratio Spy is followed successively by：110、110、111、101；The multistage smooth type at least includes：Nine rank smooth types [614000023] and ten quadravalences Smooth type [51324000000556].
Further, the fec arrangement of convolutional code provided in an embodiment of the present invention, in addition to：
Second abstraction module, for extracting the sequence to be fractionated in the convolutional code sequence；The sequence to be fractionated includes The bit of 3rd setting digit；
Processing module is classified, the sequence to be fractionated is carried out at classification for the characteristic parameter according to convolutional code sequence Reason, obtains multiple multistage sequences；Each described multistage sequence corresponds to the interference pattern of the first setting number；The interference figure Case character pair parameter has general character；The characteristic parameter at least includes：Warp equivalence value and error correction equivalence value；Described warp etc. Valid value is that the sequence of cells information obtained to the sequence progress warp processing to be fractionated is identical.
Further, the fec arrangement of convolutional code provided in an embodiment of the present invention, in addition to：
Convolutional encoding processing module, for carrying out convolutional encoding processing to the obtained maximum-likelihood decoding sequence, is obtained To covering convolutional code sequence；
Second overlay module, for utilizing the covering convolutional code sequential covering by the convolutional code sequence after noise jamming.
The embodiment of the present invention additionally provides a kind of decoder, including the decoder is used for the forward direction that performs above-mentioned convolutional code Error correction method.
The fec arrangement and decoder of convolutional code provided in an embodiment of the present invention, convolutional code is extracted using the first mechanism Sequence to be searched in sequence, and the smooth through spy of directed euler graph rule is met according to the corresponding generator matrix of convolutional code Property, detect whether sequence to be searched can accurately decode；If so, treating search sequence enters row decoding processing, and utilize what is obtained Then the covering of normal solution convolutional code is based on convolutional code by the corresponding convolutional code of convolutional code sequence after noise jamming using the second mechanism The corresponding generator matrix of strong convergence characteristic and convolutional code of sequence meets the smooth Penetrating air-flow of directed euler graph rule, from quilt Subsequent interference sequence is separated in convolutional code after covering, and enters row decoding processing to the subsequent interference sequence, maximum likelihood is obtained Coding sequence, with the decoder of the prior art based on maximum likelihood folding coding algorithm, improves the means pair of decoding performance Compared the problem of the coding process complexity and decoder of constraint length length have threshold effect in itself, it is thought based on mathematics Think, use transmission convolutional code of second mechanism to information source end in the way of error correction after first decoding to enter row decoding processing, decoding speed The fast and degree of accuracy is high；And also start the first mechanism in real time to be monitored the first mechanism, supervise the second mechanism to maintain correctly Decoding path on, effectively weaken the problem of the second mechanism steps into erroneous path, efficiently solve only start a mechanism Threshold effect problem.
The code translator that the embodiment of the present invention is provided for the specific hardware in equipment or can be installed in equipment Software or firmware etc..The technique effect and preceding method of the device that the embodiment of the present invention is provided, its realization principle and generation are real Apply example identical, to briefly describe, device embodiment part does not refer to part, refers to corresponding contents in preceding method embodiment. The technical staff in the field can be understood that, for convenience and simplicity of description, system, device and list described above The specific work process of member, may be referred to the corresponding process in above method embodiment, will not be repeated here.
, can be by others side in embodiment provided by the present invention, it should be understood that disclosed apparatus and method Formula is realized.Device embodiment described above is only schematical, for example, the division of the unit, only one kind are patrolled Collect function to divide, there can be other dividing mode when actually realizing, in another example, multiple units or component can combine or can To be integrated into another system, or some features can be ignored, or not perform.It is another, it is shown or discussed each other Coupling or direct-coupling or communication connection can be the INDIRECT COUPLING or communication link of device or unit by some communication interfaces Connect, can be electrical, machinery or other forms.
The unit illustrated as separating component can be or may not be it is physically separate, it is aobvious as unit The part shown can be or may not be physical location, you can with positioned at a place, or can also be distributed to multiple On NE.Some or all of unit therein can be selected to realize the mesh of this embodiment scheme according to the actual needs 's.
In addition, each functional unit in the embodiment that the present invention is provided can be integrated in a processing unit, also may be used To be that unit is individually physically present, can also two or more units it is integrated in a unit.
If the function is realized using in the form of SFU software functional unit and is used as independent production marketing or in use, can be with It is stored in a computer read/write memory medium.Understood based on such, technical scheme is substantially in other words The part contributed to prior art or the part of the technical scheme can be embodied in the form of software product, the meter Calculation machine software product is stored in a storage medium, including some instructions are to cause a computer equipment (can be individual People's computer, server, or network equipment etc.) perform all or part of step of each of the invention embodiment methods described. And foregoing storage medium includes：USB flash disk, mobile hard disk, read-only storage (ROM, Read-Only Memory), arbitrary access are deposited Reservoir (RAM, Random Access Memory), magnetic disc or CD etc. are various can be with the medium of store program codes.
It should be noted that：Similar label and letter represents similar terms in following accompanying drawing, therefore, once a certain Xiang Yi It is defined in individual accompanying drawing, then it further need not be defined and explained in subsequent accompanying drawing, in addition, term " the One ", " second ", " the 3rd " etc. are only used for distinguishing description, and it is not intended that indicating or implying relative importance.
Finally it should be noted that：Embodiment described above, is only the embodiment of the present invention, to illustrate the present invention Technical scheme, rather than its limitations, protection scope of the present invention is not limited thereto, although with reference to the foregoing embodiments to this hair It is bright to be described in detail, it will be understood by those within the art that：Any one skilled in the art The invention discloses technical scope in, it can still modify to the technical scheme described in previous embodiment or can be light Change is readily conceivable that, or equivalent substitution is carried out to which part technical characteristic；And these modifications, change or replacement, do not make The essence of appropriate technical solution departs from the spirit and scope of technical scheme of the embodiment of the present invention.The protection in the present invention should all be covered Within the scope of.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.

Claims (10)

1. a kind of forward error correction of convolutional code, it is characterised in that including：
Receive information source end send by the convolutional code sequence after noise jamming；The convolutional code sequence includes multiple convolution words, institute State convolution word be that the information source end is generated according to generator matrix and its carry sequence of cells information and superposition interference sequence；
Extract the sequence to be searched in the convolutional code sequence；
Meet the smooth Penetrating air-flow of directed euler graph rule according to the corresponding generator matrix of convolutional code, detect the sequence to be searched Whether row can accurately decode；The smooth Penetrating air-flow is that the convolutional code has continuity and the continuity meets generation The constraint rule of matrix setting；
If so, treating search sequence enters row decoding processing, obtained coding sequence is handled using decoding and generates normal solution convolutional code sequence Row, and utilize described in normal solution convolutional code sequential covering by the corresponding convolutional code of convolutional code sequence after noise jamming；
Directed euler graph rule is met according to the strong convergence characteristic of convolutional code sequence and the corresponding generator matrix of the convolutional code Smooth Penetrating air-flow, separate subsequent interference sequence in the convolutional code after capped；
Enter row decoding processing to the subsequent interference sequence, obtain maximum-likelihood decoding sequence.
2. the forward error correction of convolutional code according to claim 1, it is characterised in that according to the corresponding generation of convolutional code Matrix meets the smooth Penetrating air-flow of directed euler graph rule, detects whether the sequence to be searched can accurately decode, including：
The sequence to be searched is divided into multiple deconvolution time slot groups, multiple deconvolution time slot groups is met following first setting Condition；The sequence to be searched includes the bit of the first setting digit；Described first impose a condition for：The adjacent institute of any two Stating deconvolution time slot group includes the overlapping bit of the second setting digit；
Multiple deconvolution time slot groups using deconvolution formula respectively to division carry out deconvolution calculating, generation and the deconvolution The corresponding multiple coding sequences of time slot group；Each described coding sequence includes multiple warp cells；
Detect whether multiple warp cells overlapping in arbitrary neighborhood coding sequence in multiple coding sequences are equal；
If so, determining that the sequence to be searched can be decoded accurately.
3. the forward error correction of convolutional code according to claim 1, it is characterised in that according to the strong receipts of convolutional code sequence Hold back characteristic and the corresponding generator matrix of the convolutional code meets the smooth Penetrating air-flow of directed euler graph rule, after capped The convolutional code in separate subsequent interference sequence, including：
According to the maximum of the generator matrix for generating convolutional code freely away from building for convolutional code sequence described in separating treatment Multiple multistage smooth types；The smooth type includes：Unrelated bit and the related bits for convolutional code sequence described in separating treatment； The related bits include the characteristic parameter for being used to detect the convolutional code sequence：It is described that each multistage smooth type includes Related bits are different, and the characteristic parameter of different related bits detections is different；The related bits are the Europe The point in any a line that drawing figure includes；The point outside any a line that the unrelated bit then includes for the Euler diagram；
According to multiple multistage smooth types, the subsequent interference sequence and sequence of cells letter are separated from the convolutional code sequence Breath.
4. the forward error correction of convolutional code according to claim 3, it is characterised in that according to for generating convolutional code The maximum of generator matrix freely away from, build for the multiple multistage smooth of convolutional code sequence described in separating treatment, including：
According to the maximum of the corresponding generator matrix of the convolutional code freely away from the most short multistage smooth type of structure；
According to the most short multistage smooth type, multiple multistage smooth types are built, the exponent number of each multistage smooth type is big In the exponent number of the most short multistage smooth type.
5. the forward error correction of convolutional code according to claim 4, it is characterised in that the convolutional code be information source end by Generated according to the generator matrix of [3,1,8] nonsystematic convolutional code；
The most short multistage smooth type is [3375], [3375] be quadravalence smooth type and corresponding related bits and unrelated bit according to It is secondary to be：110、110、111、101；The multistage smooth type at least includes：Nine rank smooth types [614000023] and ten quadravalences are smooth Formula [51324000000556].
6. the forward error correction of convolutional code according to claim 3, it is characterised in that according to the strong receipts of convolutional code sequence Hold back characteristic and the corresponding generator matrix of the convolutional code meets the smooth Penetrating air-flow of directed euler graph rule, after capped The convolutional code in separation subsequent interference sequence before, in addition to：
Extract the sequence to be fractionated in the convolutional code sequence；The sequence to be fractionated includes the bit of the 3rd setting digit；
Classification processing is carried out to the sequence to be fractionated according to the characteristic parameter of convolutional code sequence, multiple multistage sequences are obtained；Often One multistage sequence corresponds to the interference pattern of the first setting number；The interference pattern character pair parameter has altogether Property；The characteristic parameter at least includes：Warp equivalence value and error correction equivalence value；The warp equivalence value is to the sequence to be fractionated The sequence of cells that row progress warp processing is obtained is identical.
7. the forward error correction of convolutional code according to claim 3, it is characterised in that translate the interference sequence Code processing, after obtaining maximum-likelihood decoding sequence, in addition to：
Convolutional encoding processing is carried out to the obtained maximum-likelihood decoding sequence, obtains covering convolutional code sequence；
Using the covering convolutional code sequential covering by the convolutional code sequence after noise jamming.
8. a kind of fec arrangement of convolutional code, it is characterised in that including：
Receiving module, for receive information source end transmission by the convolutional code sequence after noise jamming；The convolutional code sequence includes Multiple convolution words, the convolution word is that the information source end is generated according to generator matrix and it carries sequence of cells information and dry Disturb sequence；
Abstraction module, for extracting the sequence to be searched in the convolutional code sequence；
Detection module, the smooth Penetrating air-flow for meeting directed euler graph rule according to the corresponding generator matrix of convolutional code, inspection Survey whether the sequence to be searched can accurately decode；The smooth Penetrating air-flow is that the convolutional code has continuity and described Continuity meets the constraint rule of generator matrix setting；
Overlay module, for when the sequence to be searched can be decoded accurately, entering row decoding processing to the sequence to be searched, The normal solution convolutional code sequence that obtained coding sequence generation is obtained is handled using decoding, and it is corresponding using normal solution convolutional code sequence Normal solution convolutional code covers the corresponding convolutional code of convolutional code sequence by after noise jamming；
Separation module, meeting for the strong convergence characteristic according to convolutional code sequence and the corresponding generator matrix of the convolutional code has To the smooth Penetrating air-flow of Euler diagram rule, subsequent interference sequence is separated in the convolutional code after being capped；
Processing module is decoded, for entering row decoding processing to the subsequent interference sequence, maximum-likelihood decoding sequence is obtained.
9. the fec arrangement of convolutional code according to claim 8, it is characterised in that the detection module, including：
Division unit, for the sequence to be searched to be divided into multiple deconvolution time slot groups, expires multiple deconvolution time slot groups First is enough down to impose a condition；The sequence to be searched includes the bit of the first setting digit；Described first impose a condition for：Appoint Two adjacent deconvolution time slot groups of meaning include the overlapping bit of the second setting digit；
Generation unit, deconvolution calculating is carried out for multiple deconvolution time slot groups using deconvolution formula respectively to division, raw Into multiple coding sequences corresponding with the deconvolution time slot group；Each described coding sequence includes multiple warp cells；
Whether detection unit, multiple warp cells for detecting overlapping in arbitrary neighborhood coding sequence in multiple coding sequences are equal It is equal；
Determining unit, it is equal for overlapping in arbitrary neighborhood coding sequence in detecting multiple coding sequences multiple warp cells When equal, determine that the sequence to be searched can be decoded accurately.
10. a kind of decoder, it is characterised in that the decoder is used for the convolutional code described in perform claim requirement any one of 1-7 Forward error correction.
CN201710500396.XA 2017-06-27 2017-06-27 Forward error correction method, device and decoder of convolutional code Active CN107294544B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710500396.XA CN107294544B (en) 2017-06-27 2017-06-27 Forward error correction method, device and decoder of convolutional code

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710500396.XA CN107294544B (en) 2017-06-27 2017-06-27 Forward error correction method, device and decoder of convolutional code

Publications (2)

Publication Number Publication Date
CN107294544A true CN107294544A (en) 2017-10-24
CN107294544B CN107294544B (en) 2020-09-01

Family

ID=60098316

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710500396.XA Active CN107294544B (en) 2017-06-27 2017-06-27 Forward error correction method, device and decoder of convolutional code

Country Status (1)

Country Link
CN (1) CN107294544B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109586818A (en) * 2018-12-21 2019-04-05 北京中科汉天下电子技术有限公司 The signal-noise ratio estimation method and device of constant enveloped modulation signal

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1392669A (en) * 2002-08-12 2003-01-22 王继深 Decoder design method for real time high speed anti-interference coding
CN1607759A (en) * 2003-09-01 2005-04-20 阿尔卡特公司 Method of decoding a data word
CN101176263A (en) * 2005-05-13 2008-05-07 韦里佐内服务公司 Systems and methods for decoding forward error correcting codes
CN102624402A (en) * 2012-04-01 2012-08-01 深圳市统先科技股份有限公司 LDPC (low density parity check) decoder
CN102891737A (en) * 2012-10-18 2013-01-23 电子科技大学 Method and system for coding and decoding binary rateless codes
US20140223267A1 (en) * 2010-12-02 2014-08-07 Lsi Corporation Radix-4 viterbi forward error correction decoding
CN106603196A (en) * 2016-11-22 2017-04-26 西安电子科技大学 Convolutional network error-correcting code coding and decoding method

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1392669A (en) * 2002-08-12 2003-01-22 王继深 Decoder design method for real time high speed anti-interference coding
CN1607759A (en) * 2003-09-01 2005-04-20 阿尔卡特公司 Method of decoding a data word
CN101176263A (en) * 2005-05-13 2008-05-07 韦里佐内服务公司 Systems and methods for decoding forward error correcting codes
US20140223267A1 (en) * 2010-12-02 2014-08-07 Lsi Corporation Radix-4 viterbi forward error correction decoding
CN102624402A (en) * 2012-04-01 2012-08-01 深圳市统先科技股份有限公司 LDPC (low density parity check) decoder
CN102891737A (en) * 2012-10-18 2013-01-23 电子科技大学 Method and system for coding and decoding binary rateless codes
CN106603196A (en) * 2016-11-22 2017-04-26 西安电子科技大学 Convolutional network error-correcting code coding and decoding method

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
LEE S J, GOEL M, ZHU Y, ET AL.: "Forward error correction decoding for WiMAX and 3GPP LTE modems", 《2008 42ND ASILOMAR CONFERENCE ON SIGNALS, SYSTEMS AND COMPUTERS》 *
李小文; 罗友宝: "一种应用于LTE系统的Viterbi译码算法", 《电信科学》 *
王继深: "铁路运输管理信息系统接入数据交换网的联网实践", 《铁路通信信号设计》 *
赵冰: "卷积编码及基于DSP的Viterbi译码器设计", 《信息与控制》 *
钟志，张微微，单明广: "一种TPC硬判决译码改进算法研究", 《哈尔滨工业大学学报》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109586818A (en) * 2018-12-21 2019-04-05 北京中科汉天下电子技术有限公司 The signal-noise ratio estimation method and device of constant enveloped modulation signal
CN109586818B (en) * 2018-12-21 2021-05-07 北京昂瑞微电子技术股份有限公司 Signal-to-noise ratio estimation method and device for constant envelope modulation signal

Also Published As

Publication number Publication date
CN107294544B (en) 2020-09-01

Similar Documents

Publication Publication Date Title
CN104462253B (en) A kind of topic detection or tracking of network-oriented text big data
Vishkin Optimal parallel pattern matching in strings
Heras et al. Proof-pattern recognition and lemma discovery in ACL2
CN104025459A (en) Decoding processing method and decoder
CN107147400A (en) A kind of IRA codes opener blind-identification method
CN110492974A (en) A kind of parallel polarization code coding method and device
CN106712898A (en) Channel coding blind recognition method based on Gauss iteration column elimination
Amir et al. Cycle detection and correction
Grohe et al. Learning MSO-definable hypotheses on strings
CN107294544A (en) Forward error correction, device and the decoder of convolutional code
CN107370555A (en) Error decision method and its device
CN111726125B (en) Error-containing concatenated code identification method based on template matching
CN109542508B (en) Code homology analysis method based on function call relation graph
CN115562679B (en) Java language-based automatic code generation method and server
Stahl et al. eRules: a modular adaptive classification rule learning algorithm for data streams
US6885701B2 (en) Data demodulation apparatus and method
CN113177191A (en) Firmware function similarity detection method and system based on fuzzy matching
Manvel Reconstructing the degree pair sequence of a digraph
Chen et al. AMOS: An automated model order selection algorithm for spectral graph clustering
Fredriksson et al. Rotation and lighting invariant template matching
Bonichon et al. Short labels by traversal and jumping
CN109582930A (en) Sliding input decoding method and device and electronic equipment
Bodlaender et al. Domino treewidth
Ramesh et al. Optimal speedups for parallel pattern matching in trees
Yohananov et al. Codes over trees

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant