CN110212922A - A kind of polarization code adaptive decoding method and system - Google Patents
A kind of polarization code adaptive decoding method and system Download PDFInfo
- Publication number
- CN110212922A CN110212922A CN201910476954.2A CN201910476954A CN110212922A CN 110212922 A CN110212922 A CN 110212922A CN 201910476954 A CN201910476954 A CN 201910476954A CN 110212922 A CN110212922 A CN 110212922A
- Authority
- CN
- China
- Prior art keywords
- path
- decoding
- layer
- neural network
- unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/13—Linear codes
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L1/00—Arrangements for detecting or preventing errors in the information received
- H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
- H04L1/0045—Arrangements at the receiver end
- H04L1/0052—Realisations of complexity reduction techniques, e.g. pipelining or use of look-up tables
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L1/00—Arrangements for detecting or preventing errors in the information received
- H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
- H04L1/0056—Systems characterized by the type of code used
- H04L1/0057—Block codes
Abstract
The invention discloses a kind of polarization code adaptive decoding method and system, belong to wireless communication field.Method of the invention is first to construct neural network and be trained, then routing information to be decoded is input to neural network and obtains every layer of maximum path number；Every layer of maximum path number is then input to decoder, recycles decoder to treat decoding path information and is decoded to obtain decoding result.System of the invention includes neural network unit and decoding unit, and neural network unit and decoding unit are electrically connected；Decoding unit includes Path extension unit and sequencing unit, and Path extension unit and sequencing unit electrical connection, Path extension unit are electrically connected with neural network unit.It is an object of the invention to overcome in the prior art, the decoding algorithm training complexity based on deep learning is higher, is unsuitable for the deficiency of long code application, the present invention can reduce decoding complexity, and can satisfy communication system different channel circumstance and configuration requirement.
Description
Technical field
The present invention relates to wireless communication fields, more specifically to a kind of polarization code adaptive decoding method and system.
Background technique
So far, channel coding had been developed for more than 70 years,The polarization code of proposition, having proved to be first can be achieved
Symmetric binary inputs discrete memoryless channel (binary-input, discrete, memoryless channels, B-DMC)
The code of channel capacity.In 5G field test, polarization code achieves very big performance boost effect, and is chosen as 5G enhancing letter
Forward error correction (the forward error of road scene (enhanced mobile broadband, eMBB) control channel
Correction, FEC) code.In order to meet the low latency and high-speed demand of 5G, researchers have been made many effort and come
Design the polarization code decoder with high hardware efficiency.Currently, research can balance the efficient pole of complexity and performance well
Change code decoder be it is very urgent and it is necessary to.
The decoder one of most widely used as polarization code, successive elimination (successive cancellation, SC) are translated
Code device can show its advantage in terms of complexity, but be limited by bit by bit decoding strategy, and can not achieve satisfactory
Finite length error-correcting performance.Pass through the more multipath in every grade of extension, successive elimination list (successive
Cancellation list, SCL) and successive elimination storehouse (successive cancellation stack, SCS) decoder
It is proposed as improved SC decoder.SCL decoder can be considered as the application of binary tree breadth-first search, SCS
Decoder can be approximately Depth Priority Algorithm.Simulation result, which shows SCL and SCS decoder, can be achieved close to maximum likelihood
The performance of (maximum likelihood, ML) decoder, and there is acceptable time complexity O (LNlogN), wherein L table
Show that search width, N indicate code length.
In order to avoid performance decline, the size that the search width L of SCL needs to have moderate.In this case, SCL is translated
The time complexity of code device is still slightly higher.Therefore, search width adaptive strategy and Pruning strategy are suggested to reduce and calculate again
Miscellaneous degree.But these strategies are controlled by threshold value mostly, not can guarantee the reasonability of route deletion, therefore less have performance gain.
In recent years, deep learning (deep learning, DL) causes entirely because of the ability of its powerful solution complex task
The concern in the world.By deep learning, many fields realize significant performance boost, such as computer vision, game and life
Object informatics, deep learning equally also can be applied to polarization code decoding field.Tradition is passed through based on the decoder of deep learning
Learn a large amount of code words to obtain close to the optimal bit error rate (bit error ratio, BER) performance.But it in training and runs
The complexity of exponential increase hinders its practical application in long code decoding in terms of neural network.Therefore, huge training burden
It is two major obstacles hindered using deep learning as polarization code decoding subplan with higher operation complexity.
In conclusion by observation decoding algorithm it is not difficult to find that existing SCL algorithm there are following problems: 1) it is traditional
SCL/SCS decoding algorithm search width L be it is fixed, cannot according to decoding scene and change, complexity is higher；2) it is based on path
The SCL/SCS decoding algorithm of deletion can bring certain performance loss；3) the SC decoding algorithm training complexity based on deep learning
It is higher, it is unsuitable for long code application.Based on above analysis it is found that existing polarization code coding method is not enough to meet reality well
The demand of border application.
Summary of the invention
1. to solve the problems, such as
It is an object of the invention to overcome in the prior art, SC decoding algorithm based on deep learning training complexity compared with
Height is unsuitable for the deficiency of long code application, provides a kind of polarization code adaptive decoding method, can reduce decoding complexity, and
It can satisfy communication system different channel circumstance and configuration requirement.
2. technical solution
To solve the above-mentioned problems, the technical solution adopted in the present invention is as follows:
A kind of polarization code adaptive decoding method of the invention first constructs neural network and is trained, then will be to be decoded
Routing information is input to neural network and obtains the maximum path number of every layer of information code binary tree；Then by every layer of maximum path number
It is input to decoder, decoding path is treated further according to every layer of maximum path number and is extended to obtain path candidate parallel, then
It calculates the transition probability of path candidate and is ranked up, obtain decoding result further according to path length and maximum transition probability.
Further, comprising the following steps: Step 1: building neural network, constructs a neural network and instructed
Practice, wherein neural network includes input layer, hidden layer and output layer；Step 2: prediction, routing information y to be decoded is input to
Neural network obtains every layer of maximum path number Zi, then by every layer of maximum path number ZiIt is input to decoder；Wherein, i ∈
(1 ..., K-2), K are the number of information bit；Step 3: decoder initializes, enabling the width of Q in decoder is Lmax, Q is storage
The list in every layer of path；Again by Q, Q decoding path and storage matrix be initialized as sky；Wherein, the decoding path in Q is(i∈(1,…,N)；Step 4: extension, in the information code binary tree of decoder, if information code
The i-stage of binary tree is to freeze position, then is extended to decoding paths all in QThe decoding path after extension is deposited again
Storage is into Q and executes step 6；If the i-stage of information code binary tree is information bit, this layer of corresponding maximum path number is read
Zi, and by Z preceding in QiThe path candidate that paths extend isAndAnd the transition probability of path candidate
It is different；Step 5: sequence, calculates the transition probability of path candidate and is ranked up, then descending selection LmaxPath candidate
It deposits into Q；Step 6: resolution exports the maximum path of Q transition probability if the path length in Q is identical as code length N
As decoding as a result, being otherwise back to step 4.
Further, every layer of maximum path number is calculated according to the following formula:
Wherein, WijFor every layer of corresponding weight, bjFor every layer of corresponding biasing, xjFor every layer of corresponding input value.
Further, the calculation formula of transition probability is as follows:
Wherein,Indicate that input is u1Shi Changdu is the transition probability in the path of i,Indicate length
For the odd term of the decoding path of i,Indicate that length is the even item of the decoding path of i,For the codeword vector received,For previous decoding path, uiIt is inputted for previous channel.
Further, the nodal point number of input layer is equal to code length N, and input vector is routing information y to be decoded, hidden layer
Nodal point number is 8*N, and output layer nodal point number is K -2.
Further, neural network is trained using TensorFlow platform, wherein maximum number of iterations Tepoch
It is 50.
A kind of decoding system using a kind of above-mentioned polarization code adaptive decoding method of the present invention, including neural network list
Member and decoding unit, neural network unit and decoding unit electrical connection；Wherein, neural network unit is for calculating information code y-bend
Every layer of maximum number of path in tree, decoding unit are decoded for treating decoding path information.
Further, decoding unit includes Path extension unit and sequencing unit, Path extension unit and sequencing unit
Electrical connection, and Path extension unit is electrically connected with neural network unit.
3. beneficial effect
Compared with the prior art, the invention has the benefit that
(1) a kind of polarization code adaptive decoding method of the invention, every layer of maximum path number is obtained by neural network,
It so as to select to determine every layer of search width, and then can be adapted for various signal-to-noise ratio scenes, improve the suitable of this method
The property used；And decoding path is treated by every layer of maximum path number and is extended to obtain path candidate parallel, to reduce every
The Path extension number of layer, thereby reduces the Time & Space Complexity of decoder；
(2) a kind of polarization code adaptive decoding system of the invention, by the way that neural network unit and decoding unit is arranged, from
And be combined together deep learning with conventional polar code decoder, it is extended by every grade of search width of prediction come constrained Path,
Therefore computation complexity is reduced, meanwhile, early interrupt of erroneous path promotes decoding performance.
Detailed description of the invention
Fig. 1 is a kind of flow diagram of polarization code adaptive decoding method of the invention；
Fig. 2 is a kind of structural schematic diagram of polarization code adaptive decoding system of the present invention；
Fig. 3 is the structural schematic diagram of 2 decoding system of embodiment；
Fig. 4 is the performance comparison schematic diagram of 2 decoding system of embodiment and tradition SC and SCL.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments；Moreover, be not between each embodiment it is relatively independent, according to
It needs can be combined with each other, to reach more preferably effect.Therefore, below to the embodiment of the present invention provided in the accompanying drawings
Detailed description is not intended to limit the range of claimed invention, but is merely representative of selected embodiment of the invention.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
His embodiment, shall fall within the protection scope of the present invention.
To further appreciate that the contents of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments.
Embodiment 1
As shown in connection with fig. 1, a kind of polarization code adaptive decoding method of the invention first constructs neural network and is instructed
Practice, then routing information to be decoded is input to neural network and obtains the maximum path number of every layer of information code binary tree；Pass through nerve
Network obtains every layer in the information code binary tree of decoder of maximum path number, so as to select the search for determining every layer wide
Degree, and then can be adapted for various signal-to-noise ratio scenes, improve the applicability of this method.Further, by every layer of most main road
Diameter number is input to decoder, treats decoding path further according to every layer of maximum path number and is extended to obtain path candidate parallel,
To reduce every layer of Path extension number, the Time & Space Complexity of decoder is thereby reduced；Then it calculates candidate
The transition probability in path is simultaneously ranked up, and obtains decoding result further according to path length and maximum transition probability.
It should be noted that routing information to be decoded is y in the present invention, code length (length of decoding) is N, information bit number
For K, every layer of maximum search width is L during decodingmax, the list in every layer of path of storage is Q, and being stored in length in Q is i
Path be(i∈(1,…,N),j∈(1,…,Lmax)), the corresponding transition probability in path is
A kind of polarization code adaptive decoding method of the invention, the specific steps are as follows:
Step 1: building neural network
It constructs a neural network and is trained, wherein neural network includes input layer, hidden layer and output layer；Value
It must illustrate, the nodal point number of input layer is equal to code length N, and input vector is routing information y to be decoded, and hidden layer nodal point number is
8*N, output layer nodal point number are K -2；The present embodiment is trained neural network using TensorFlow platform, is carrying out nerve
When network unit training, the code word that training set is generated under different signal-to-noise ratio by 240000 groups is formed, and every 120 are a batch, is learned
Habit rate is set as 0.001, maximum number of iterations TepochIt is 50.
Step 2: prediction
Routing information y to be decoded is input to neural network and obtains every layer of maximum path number Zi, then by every layer of maximum
Number of path ZiIt is input to decoder；Wherein, i ∈ (1 ..., K-2), K are the number of information bit；It is worth noting that according to following
Formula calculates every layer of maximum path number:
Wherein, WijFor weight, bjBiasing, xjFor every layer of corresponding input value, σ is activation primitive sigmoid,Wherein, a=∑jWijxj+bj。
Step 3: decoder initializes
The width for enabling Q in decoder is Lmax, Q is the list for storing every layer of path；Again by Q, Q decoding path and
Storage matrix is initialized as sky；Wherein, the decoding path in Q is(i∈(1,…,N)；
Step 4: extension
In the information code binary tree of decoder, if the i-stage of information code binary tree is to freeze position, translated all in Q
Code Path extension beThe transition probability in path is constant, then the decoding path after extension is stored into Q and executes step
Rapid six；If the i-stage of information code binary tree is information bit, this layer of corresponding maximum path number Z is readi, and by Z preceding in QiItem
The path candidate that Path extension obtains isAndAnd the transition probability of path candidate is different；
Step 5: sequence
It calculates the transition probability of path candidate and is ranked up, then descending selection LmaxPath candidate is deposited into Q；
Wherein, the calculation formula of transition probability is as follows:
Wherein,Indicate that input is u1Shi Changdu is the transition probability in the path of i,Indicate length
For the odd term of the decoding path of i,Indicate that length is the even item of the decoding path of i,For the codeword vector received,For previous decoding path, uiIt is inputted for previous channel.
The present embodiment is ranked up using transition probability of the quicksort strategy to path candidate.It is worth noting that logical
Cross descending selection LmaxPath candidate is deposited into Q, is deleted erroneous path in advance and is realized to decode better than traditional list and calculates
The performance of method.
Step 6: resolution
If the path length in Q is identical as code length, by the output of Q transition probability maximum path as decoding as a result,
Otherwise step 4 is back to continue to decode.
A kind of polarization code adaptive decoding method of the invention, by modify the parameter of neural network, scale and input to
Amount can predict the search width under different code length and code rate, can satisfy the different channel circumstance of communication system and configuration is wanted
It asks, there is flexibility ratio abundant, and there is configurability；The present invention can effectively improve at the communication performance and data of system
Reason ability, and decoding complexity can be reduced.
As shown in connection with fig. 2, a kind of polarization code adaptive decoding system of the invention, the system use a kind of above-mentioned polarization
Code self-adapting interpretation method, the system include decoding unit and neural network unit, and decoding unit and neural network unit are electrically connected
It connects, wherein neural network unit is translated for calculating every layer of maximum number of path in information code binary tree, decoding unit for treating
Code routing information is decoded.Specifically, neural network unit calculates every in information code binary tree according to routing information to be decoded
The maximum number of path of layer；It is worth noting that decoding unit is above-mentioned decoder, neural network unit of the invention includes defeated
Enter layer, hidden layer and output layer；The nodal point number of input layer is equal to code length N, and input vector is routing information y to be decoded, is hidden
Layer nodal point number is 8*N, and output layer nodal point number is K -2.
Further, decoding unit includes Path extension unit and sequencing unit, Path extension unit and neural network list
Member electrical connection, sequencing unit and the electrical connection of Path extension unit；Wherein, Path extension unit be used for read stored in Q wait translate
Code path, and treat decoding path and be extended parallel and calculate corresponding transition probability, calculated result is then input to row
Sequence unit；Sequencing unit sorts from large to small transition probability, then descending selection LmaxPath candidate is deposited into Q；
Remaining path and its corresponding transition probability are deleted, Path extension unit is returned to.A kind of polarization code self-adapting of the invention is translated
Code system by every grade of search width of prediction come constrained Path extension, therefore reduces computation complexity, meanwhile, erroneous path
Early interrupt promotes decoding performance.
Embodiment 2
Decoding system as shown in connection with fig. 3, code length N are 8, and information bit number K is 4, search width LmaxIt is 4, black node
Node is accessed by this system, grey node accesses node by traditional list decoding algorithm.The neural network list of the present embodiment
The input vector of member is the reception code word that length is 8, exports the search width sequence for being 2 for length, and be transmitted to decoding unit.
In decoding unit, as i=7, corresponding Z1When=1, therefore only a paths are extended, i=8 similarly.This system
Access number of network nodes has significant decrease compared to traditional list decoding algorithm, has significant advantage in terms of the computation complexity.
As shown in connection with fig. 4, the code length N=64 in Fig. 4, initial maximum search width Lmax=4, signal-to-noise ratio EbN0=1.0-
4.0dB, under various signal-to-noise ratio, decoding system performance of the invention is better than traditional SC and SCL decoder, and in high s/n ratio
Under, decoding system performance of the invention and theoretical optimal SCL decoder are close, and decoding system of the invention has excellent property
Energy.
Table 1 presents decoding system of the invention compared with the search width of tradition SCL decoder, wherein code length N=64,
Initial maximum search width Lmax=4, signal-to-noise ratio EbN0=1.0-4.0dB, due to the computation complexity of list polarization code decoder
It is positively correlated with Path extension number, therefore the computation complexity of design can be embodied by Path extension number, seen from table 1,
Under various signal-to-noise ratio, decoding system complexity of the invention is substantially reduced.Being computed can obtain, when signal-to-noise ratio is 4.0dB, this hair
The computation complexity of bright decoding system has dropped 67.51% compared with traditional SCL decoding algorithm, it follows that of the invention
Decoding system significantly reduces the computation complexity of algorithm, especially suitable for compared with high s/n ratio situation.
Path extension number under the different signal-to-noise ratio of table 1 compares
| EbN0(dB) | 1.0 | 2.0 | 3.0 | 4.0 |
| Traditional list decoder | 4 | 4 | 4 | 4 |
| Decoding system of the present invention | 1.6723 | 1.4987 | 1.3420 | 1.2998 |
In conclusion a kind of polarization code adaptive decoding method and system of the invention, by deep learning and conventional polar
Code decoder is combined together, and by every grade of search width of prediction come constrained Path extension, therefore reduces computation complexity, together
When, early interrupt of erroneous path promotes decoding performance.The present embodiment presents its advantage in complexity and performance,
Its high hardware adaptation also shows the great potential for practical application.
The present invention is described in detail above in conjunction with specific exemplary embodiment.It is understood, however, that can not take off
It is carry out various modifications in the case where from the scope of the present invention being defined by the following claims and modification.Detailed description and drawings
Should be to be considered only as it is illustrative and not restrictive, if there is any such modifications and variations, then they all will
It falls into the scope of the present invention described herein.In addition, Development Status and meaning that background technique is intended in order to illustrate this technology,
It is not intended to limit the present invention or the application and application field of the invention.
Claims (8)
1. a kind of polarization code adaptive decoding method, which is characterized in that first construct neural network and be trained, then will be to be decoded
Routing information is input to neural network and obtains the maximum path number of every layer of information code binary tree；Then by every layer of maximum path number
It is input to decoder, decoding path is treated further according to every layer of maximum path number and is extended to obtain path candidate parallel, then
It calculates the transition probability of path candidate and is ranked up, obtain decoding result further according to path length and maximum transition probability.
2. a kind of polarization code adaptive decoding method according to claim 1, which comprises the following steps:
Step 1: building neural network
It constructs a neural network and is trained, wherein neural network includes input layer, hidden layer and output layer；
Step 2: prediction
Routing information y to be decoded is input to neural network and obtains every layer of maximum path number Zi, then by every layer of maximum path number
ZiIt is input to decoder；Wherein, i ∈ (1 ..., K-2), K are the number of information bit；
Step 3: decoder initializes
The width for enabling Q in decoder is Lmax, Q is the list for storing every layer of path；Again by Q, Q decoding path and storage
Matrix initialisation is sky；Wherein, the decoding path in Q is
Step 4: extension
In the information code binary tree of decoder, if the i-stage of information code binary tree is to freeze position, by decodings all in Q road
Diameter is extended toThe decoding path after extension is stored into Q again and executes step 6；If the i-th of information code binary tree
Grade is information bit, then reads this layer of corresponding maximum path number Zi, and by Z preceding in QiThe path candidate that paths extend isAndAnd the transition probability of path candidate is different；
Step 5: sequence
It calculates the transition probability of path candidate and is ranked up, then descending selection LmaxPath candidate is deposited into Q；
Step 6: resolution
If the path length in Q is identical as code length N, by the maximum path output of Q transition probability as decoding as a result, otherwise
It is back to step 4.
3. a kind of polarization code adaptive decoding method according to claim 2, which is characterized in that calculate according to the following formula
Every layer of maximum path number:
Wherein, WijFor weight, bjFor biasing, xjFor every layer of corresponding input value.
4. a kind of polarization code adaptive decoding method according to claim 2, which is characterized in that the calculating of transition probability is public
Formula is as follows:
Wherein,Indicate that input is u1Shi Changdu is the transition probability in the path of i,Indicate that length is i's
The odd term of decoding path,Indicate that length is the even item of the decoding path of i,For the codeword vector received,For
Previous decoding path, uiIt is inputted for previous channel.
5. a kind of polarization code adaptive decoding method according to claim 2, which is characterized in that the nodal point number etc. of input layer
In code length N, and input vector is routing information y to be decoded, and hidden layer nodal point number is 8*N, and output layer nodal point number is K -2.
6. a kind of described in any item polarization code adaptive decoding methods according to claim 1~5, which is characterized in that utilize
TensorFlow platform is trained neural network, wherein maximum number of iterations TepochIt is 50.
7. it is a kind of using a kind of described in any item decoding systems of polarization code adaptive decoding method of claim 1~6, it is special
Sign is, including neural network unit and decoding unit, the neural network unit and decoding unit electrical connection；Wherein, neural
Network unit for calculating every layer of maximum number of path in information code binary tree, decoding unit for treat decoding path information into
Row decoding.
8. a kind of polarization code adaptive decoding system according to claim 7, which is characterized in that decoding unit includes path
Expanding element and sequencing unit, the Path extension unit and sequencing unit electrical connection, and Path extension unit and neural network
Unit electrical connection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910476954.2A CN110212922B (en) | 2019-06-03 | 2019-06-03 | Polarization code self-adaptive decoding method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910476954.2A CN110212922B (en) | 2019-06-03 | 2019-06-03 | Polarization code self-adaptive decoding method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110212922A true CN110212922A (en) | 2019-09-06 |
| CN110212922B CN110212922B (en) | 2022-11-11 |
Family
ID=67790444
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910476954.2A Active CN110212922B (en) | 2019-06-03 | 2019-06-03 | Polarization code self-adaptive decoding method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110212922B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110798228A (en) * | 2019-10-29 | 2020-02-14 | 南京宁麒智能计算芯片研究院有限公司 | Polarization code turning decoding method and system based on deep learning |
| CN112332863A (en) * | 2020-10-27 | 2021-02-05 | 东方红卫星移动通信有限公司 | Polar code decoding algorithm, receiving end and system under low signal-to-noise ratio scene of low earth orbit satellite |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016168962A1 (en) * | 2015-04-20 | 2016-10-27 | 华为技术有限公司 | Decoding method and decoding apparatus for polar code |
| US20170076199A1 (en) * | 2015-09-14 | 2017-03-16 | National Institute Of Information And Communications Technology | Neural network system, and computer-implemented method of generating training data for the neural network |
| CN108777584A (en) * | 2018-07-06 | 2018-11-09 | 中国石油大学(华东) | A kind of fast Optimization of polarization code decoding parameter |
| CN109450456A (en) * | 2018-10-30 | 2019-03-08 | 南京大学 | A kind of adaptive storehouse interpretation method and system based on polarization code |
-
2019
- 2019-06-03 CN CN201910476954.2A patent/CN110212922B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016168962A1 (en) * | 2015-04-20 | 2016-10-27 | 华为技术有限公司 | Decoding method and decoding apparatus for polar code |
| US20170076199A1 (en) * | 2015-09-14 | 2017-03-16 | National Institute Of Information And Communications Technology | Neural network system, and computer-implemented method of generating training data for the neural network |
| CN108777584A (en) * | 2018-07-06 | 2018-11-09 | 中国石油大学(华东) | A kind of fast Optimization of polarization code decoding parameter |
| CN109450456A (en) * | 2018-10-30 | 2019-03-08 | 南京大学 | A kind of adaptive storehouse interpretation method and system based on polarization code |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110798228A (en) * | 2019-10-29 | 2020-02-14 | 南京宁麒智能计算芯片研究院有限公司 | Polarization code turning decoding method and system based on deep learning |
| CN112332863A (en) * | 2020-10-27 | 2021-02-05 | 东方红卫星移动通信有限公司 | Polar code decoding algorithm, receiving end and system under low signal-to-noise ratio scene of low earth orbit satellite |
| CN112332863B (en) * | 2020-10-27 | 2023-09-05 | 东方红卫星移动通信有限公司 | Polar code decoding algorithm, receiving end and system under low signal-to-noise ratio scene of low orbit satellite |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110212922B (en) | 2022-11-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110322446B (en) | Domain self-adaptive semantic segmentation method based on similarity space alignment | |
| CN111292808B (en) | DNA storage coding optimization method based on improved Harris eagle algorithm | |
| CN110533096B (en) | DNA storage coding optimization method of multivariate universe algorithm based on K-means clustering | |
| CN109165720A (en) | Neural network model compression method, device and computer equipment | |
| CN110212922A (en) | A kind of polarization code adaptive decoding method and system | |
| CN110233628B (en) | Self-adaptive belief propagation list decoding method for polarization code | |
| CN107392919A (en) | Gray threshold acquisition methods, image partition method based on self-adapted genetic algorithm | |
| CN106849960A (en) | Segmentation CRC check storehouse interpretation method and framework based on polarization code | |
| CN109450459B (en) | Polarization code FNSC decoder based on deep learning | |
| CN110362683A (en) | A kind of information steganography method based on recurrent neural network, device and storage medium | |
| Ye et al. | Towards quantifiable dialogue coherence evaluation | |
| CN113885329A (en) | Mobile robot path planning method based on deep reinforcement learning | |
| CN112332864A (en) | Polar code decoding method and system for self-adaptive ordered mobile pruning list | |
| CN114070331B (en) | Self-adaptive serial offset list flip decoding method and system | |
| CN109450456A (en) | A kind of adaptive storehouse interpretation method and system based on polarization code | |
| CN108964672A (en) | A kind of polarization code coding method based on deep neural network | |
| CN111339635B (en) | DNA storage coding optimization method of multi-element universe algorithm based on damping factors | |
| CN105376185A (en) | Constant modulus blind equalization processing method based on optimization of DNA shuffled frog leaping algorithm in communication system | |
| Zhang et al. | A hybrid sphere decoding for short polar codes using variable step size | |
| CN111697975A (en) | Polarization code continuous deletion decoding optimization algorithm based on full-connection neural network | |
| CN115828154B (en) | LPI radar signal identification method, system, equipment and storage medium | |
| CN113114423B (en) | Adaptive sphere decoding detection method | |
| CN117478151A (en) | SCL decoding path deleting strategy based on neural network | |
| CN116842730A (en) | DNA storage coding set construction method based on improved firefly search algorithm | |
| CN113965209A (en) | Self-adaptive SCL decoding algorithm based on two-classification convolutional network |
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 |