CN109428673A - For the method for decoded signal, equipment and storage equipment - Google Patents
For the method for decoded signal, equipment and storage equipment Download PDFInfo
- Publication number
- CN109428673A CN109428673A CN201710753622.5A CN201710753622A CN109428673A CN 109428673 A CN109428673 A CN 109428673A CN 201710753622 A CN201710753622 A CN 201710753622A CN 109428673 A CN109428673 A CN 109428673A
- Authority
- CN
- China
- Prior art keywords
- noise
- estimation
- signal
- interchannel noise
- interchannel
- 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.)
- Pending
Links
Classifications
-
- 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
- H04L1/0047—Decoding adapted to other signal detection operation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03165—Arrangements for removing intersymbol interference using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Power Engineering (AREA)
- Noise Elimination (AREA)
Abstract
The embodiment of the present disclosure provides a kind of method for decoded signal.This method comprises: being decoded using predetermined decoding scheme to signal is received, and the estimation to signal is sent is obtained based on decoding result；The estimation to interchannel noise is obtained using the correlation of interchannel noise；The estimation obtained to interchannel noise is subtracted from the reception signal, to obtain the reception signal of modification；And be decoded using reception signal of the predetermined decoding scheme to the modification, to obtain decoded signal.The embodiment of the present disclosure additionally provides the equipment and storage equipment for decoded signal.
Description
Technical field
This disclosure relates to which signal decodes, and in particular, to for the method for decoded signal, equipment and storage equipment.
Background technique
Since Shannon creating information opinion, channel coding/decoding field achieves significant progress.By means of appropriate volume
Code design and efficient confidence spread decode (belief propagation, BP) algorithm, low-density checksum
(LDPC, low-density parity-check) code can obtain the performance close to shannon limit.But traditional encoding and decoding are set
Meter is primarily directed to such as Gaussian white noise channel channel design simple in this way.However this codec design does not consider
It is (this to make an uproar because of the case where the Complex Channel caused by the presence of the factors such as filtering, over-sampling, multi-user interference into actual channel
Sound is referred to as coloured noise).
Therefore, it is necessary to a kind of decoding schemes for being capable of handling this coloured noise.
Summary of the invention
The one aspect of the embodiment of the present disclosure provides a kind of method for decoded signal.This method comprises: using pre-
Definite decoding scheme is decoded to signal is received, and obtains the estimation to signal is sent based on decoding result；Utilize interchannel noise
Correlation obtain the estimation to interchannel noise；The estimation obtained to interchannel noise is subtracted from the reception signal,
To obtain the reception signal of modification；And be decoded using reception signal of the predetermined decoding scheme to the modification, with
Obtain decoded signal.
Optionally, the estimation to interchannel noise is obtained using the correlation of interchannel noise can include: by connecing from described
The estimation of described pair of transmission signal is subtracted in the collection of letters number to obtain the first estimation to interchannel noise；And utilize interchannel noise
Correlation to obtained handle the first of interchannel noise the estimation obtains the second estimation to interchannel noise, with work
For the estimation to interchannel noise.
Optionally, using the correlation of interchannel noise come to it is obtained to the first of interchannel noise the estimation handled come
Obtain the second estimation to interchannel noise can include: by first estimation to interchannel noise as the defeated of convolutional neural networks
Enter, using the output of the convolutional neural networks as second estimation to interchannel noise.
Optionally, this method, which may also include using predetermined policy, trains the convolutional neural networks, wherein described predetermined
Strategy includes any one of following: minimizing the Training strategy of surplus noise power, and is minimizing the same of surplus noise power
When make the distribution of surplus noise close to the Training strategy of Gaussian Profile.
Optionally, this method, which may also include using the reception signal of the modification as signal is received, carrys out iteration execution reception letter
Number decoding, the acquisition based on the correlation of interchannel noise to the reception signal of the estimation of interchannel noise and modification, until being
System state meets pre-provisioning request.
The embodiment of the present disclosure another aspect provides a kind of equipment for decoded signal.The equipment includes decoding
Device sends signal estimator, noise estimator and modification signal generator.Decoder is used for using predetermined decoding scheme to reception
Signal is decoded.Signal estimator is sent to be used to obtain the estimation to signal is sent based on decoding result.Noise estimator is used
The estimation to interchannel noise is obtained in the correlation using interchannel noise.Signal generator is modified to be used for from the reception signal
In subtract the estimation obtained to interchannel noise, with obtain modification reception signal.The decoder is also using described predetermined
Decoding scheme is decoded the reception signal of the modification, to obtain decoded signal.
Optionally, the noise estimator may include the first noise estimation module and the second noise estimation module.First makes an uproar
Sound estimation module can be used for obtaining by subtracting the estimation of described pair of transmission signal from the reception signal to interchannel noise
First estimation.Second noise estimation module can be used for using the correlation of interchannel noise come to obtained to interchannel noise
First estimation handle to obtain and estimate the second of interchannel noise, using as the estimation to interchannel noise.
Optionally, the second noise estimation module is realized using convolutional neural networks.In the case, described
Two noise estimation modules can also be used in: first to interchannel noise is estimated into the input as the convolutional neural networks,
Using the output of the convolutional neural networks as second estimation to interchannel noise.
Optionally, which may also include network training device, for training the convolutional Neural net using predetermined policy
Network, wherein the predetermined policy includes any one of the following terms: minimizing the Training strategy of surplus noise power, and
Make the distribution of surplus noise close to the Training strategy of Gaussian Profile while minimizing surplus noise power.
Optionally, in the device, using the reception signal of the modification as signal is received, it is iteratively performed the decoding
Device, the operation for sending signal estimator, the noise estimator and the modification signal generator, until system mode accords with
Close pre-provisioning request.
The embodiment of the present disclosure another aspect provides a kind of equipment for decoded signal.The equipment includes memory
And processor.Memory is for storing executable instruction.Processor is for executing the executable instruction stored in memory, to hold
The row above method.
The embodiment of the present disclosure another aspect provides one kind to carry the memory devices of computer program thereon, when
When executing the computer program by processor, the computer program makes the processor execute the above method.
Detailed description of the invention
For a more complete understanding of the present invention and its advantage, referring now to being described below in conjunction with attached drawing, in which:
Fig. 1 diagrammatically illustrates the outline flowchart of the method for decoded signal according to the embodiment of the present disclosure；
Fig. 2 diagrammatically illustrates the brief block diagram of the equipment for decoded signal according to the embodiment of the present disclosure；
Fig. 3 shows the simplified diagram of the decoding scheme according to the embodiment of the present disclosure；
The schematic diagram of the CNN structure according to used in the embodiment of the present disclosure is shown in Fig. 4；
Fig. 5 to Fig. 8 respectively illustrates the performance comparison schematic diagram for two different channels correlation models；And
Fig. 9 diagrammatically illustrates the brief block diagram of the electronic equipment according to the embodiment of the present disclosure.
Specific embodiment
Hereinafter, will be described with reference to the accompanying drawings embodiment of the disclosure.However, it should be understood that these descriptions are only exemplary
, and it is not intended to limit the scope of the present disclosure.In addition, in the following description, descriptions of well-known structures and technologies are omitted, with
Avoid unnecessarily obscuring the concept of the disclosure.
Term as used herein is not intended to limit the disclosure just for the sake of description specific embodiment.Used here as
Word " one ", " one (kind) " and "the" etc. also should include " multiple ", " a variety of " the meaning, unless in addition context clearly refers to
Out.In addition, the terms "include", "comprise" as used herein etc. show the presence of the feature, step, operation and/or component,
But it is not excluded that in the presence of or add other one or more features, step, operation or component.
There are all terms (including technical and scientific term) as used herein those skilled in the art to be generally understood
Meaning, unless otherwise defined.It should be noted that term used herein should be interpreted that with consistent with the context of this specification
Meaning, without that should be explained with idealization or excessively mechanical mode.
Shown in the drawings of some block diagrams and/or flow chart.It should be understood that some sides in block diagram and/or flow chart
Frame or combinations thereof can be realized by computer program instructions.These computer program instructions can be supplied to general purpose computer,
The processor of special purpose computer or other programmable data processing units, so that these instructions are when executed by this processor can be with
Creation is for realizing function/operation device illustrated in these block diagrams and/or flow chart.
Therefore, the technology of the disclosure can be realized in the form of hardware and/or software (including firmware, microcode etc.).Separately
Outside, the technology of the disclosure can take the form of the computer program product on the computer-readable medium for being stored with instruction, should
Computer program product uses for instruction execution system or instruction execution system is combined to use.In the context of the disclosure
In, computer-readable medium, which can be, can include, store, transmitting, propagating or transmitting the arbitrary medium of instruction.For example, calculating
Machine readable medium can include but is not limited to electricity, magnetic, optical, electromagnetic, infrared or semiconductor system, device, device or propagation medium.
The specific example of computer-readable medium includes: magnetic memory apparatus, such as tape or hard disk (HDD)；Light storage device, such as CD
(CD-ROM)；Memory, such as random access memory (RAM) or flash memory；And/or wire/wireless communication link.
Fig. 1 diagrammatically illustrates the outline flowchart of the method for decoded signal according to the embodiment of the present disclosure.
As shown in Figure 1, this method includes operation S110, it is decoded using predetermined decoding scheme to signal is received, and base
The estimation to signal is sent is obtained in decoding result.
Predetermined decoding scheme referred to herein can be applicable any decoding scheme, such as can be confidence spread solution
Code (belief propagation, BP) scheme according to specific application scenarios, is also possible to be used with sending side certainly
The corresponding any decoding scheme of encoding scheme.
In some instances, it obtains to can be the estimation for sending signal based on decoding result and utilize and used decoding
The corresponding encoding scheme of scheme recompiles decoding result, will recompile encoded signal obtained and be used as to hair
The estimation for the number of delivering letters.Certainly, can also can be used for estimating in receiving side sending signal using other in this field are any herein
Technical solution, the embodiment of the present disclosure are not limited by the specific implementation that signal is estimated is sent.
In operation s 120, the estimation to interchannel noise is obtained using the correlation of interchannel noise.
By in view of in channel because of the correlation of the noise caused by the factors such as filtering, over-sampling, multi-user interference,
In the embodiment of the present disclosure noise in actual channel can be more accurately estimated using the correlation.
In some instances, the estimation to interchannel noise is obtained using the correlation of interchannel noise can include: by from
Receive the first estimation for being subtracted in signal to the estimation for sending signal and being obtained to interchannel noise；And the phase using interchannel noise
Guan Xinglai is obtained and is estimated the second of interchannel noise to obtained handle the first of interchannel noise the estimation, using as
Operate the estimation described in S120 to interchannel noise.
In some instances, convolutional neural networks (CNN, convolutional neural network) can be used to come in fact
Now to the utilization of the correlation of interchannel noise.In the case, convolutional Neural net will can be used as to the first estimation of interchannel noise
The input of network, using the output of convolutional neural networks as the second estimation to interchannel noise.
Before being obtained using convolutional neural networks to the second estimation of interchannel noise, using predetermined policy to convolution
Neural network is trained, so that convolutional neural networks can simulate the scene of realistic channels.In some instances, this is predetermined
Strategy may include the Training strategy for minimizing surplus noise power, such as baseline BP-CNN (Baseline BP-CNN) strategy.?
In other examples, predetermined policy, which also is included in while minimizing surplus noise power, makes the distribution of surplus noise close to height
The Training strategy of this distribution, for example, enhancing BP-CNN (Enhanced BP-CNN) strategy.Certainly, it can be used in the embodiment of the present disclosure
It is not limited to both the above in the strategy of training convolutional neural networks, but may include the convolutional neural networks that can enable after training
Any other strategy of enough channels that more accurately reflects reality.
After obtaining the estimation to interchannel noise using the correlation of interchannel noise in operation s 120, in operation S130
In, the estimation obtained to interchannel noise is subtracted from receiving in signal, to obtain the reception signal of modification.
Due to obtaining the good estimation to interchannel noise using the correlation of interchannel noise in operation s 120, by
It receives and subtracts the good estimation in signal, the reception signal of modification obtained can have better access to the hair sent at sending side
The number of delivering letters.
In in some cases, being only performed once operation S120 may not be obtained to actual channel noise (for example, having
Coloured noise) good estimation.It is connect in this case, it is possible to which the reception signal of above-mentioned modification is carried out iteration execution as reception signal
The decoding (S110) of the collection of letters number is believed based on reception of the correlation of interchannel noise to the estimation (S120) of interchannel noise and modification
Number acquisition (S130), until system mode meets pre-provisioning request.Pre-provisioning request described herein can be time of system iteration
Number reaches pre-set number or decoded signal no longer changes, or is also possible to those skilled in the art commonly used in noise
Other any requirements of estimation.
Then, in operation S140, the reception of the modification is believed using with identical predetermined decoding scheme in operation S110
It number is decoded to obtain decoded signal.
In method shown in Fig. 1, the good estimation to interchannel noise is obtained by the correlation using interchannel noise
And the good estimation is subtracted in receiving signal, the estimation to transmission signal can be relatively accurately obtained, and realize in turn
The decoding of high quality.
Fig. 2 diagrammatically illustrates the brief block diagram of the equipment for decoded signal according to the embodiment of the present disclosure.Shown in Fig. 2
Block diagram correspond to Fig. 1 shown in method flow chart.It should be noted that for clarity and brevity, block diagram shown in Fig. 2 only shows
Function/module of the embodiment of the present disclosure is gone out to help to understand.In concrete implementation, it may also comprise more or fewer function
Energy/module.
As shown in Fig. 2, the equipment includes decoder 210, sends signal estimator 220, noise estimator 230 and modification letter
Number generator 240.
Decoder 210 is used to be decoded using predetermined decoding scheme to signal is received.
Predetermined decoding scheme referred to herein can be applicable any decoding scheme, such as can be confidence spread solution
Code (belief propagation, BP) scheme according to specific application scenarios, is also possible to be used with sending side certainly
The corresponding any decoding scheme of encoding scheme.
Signal estimator 220 is sent to be used to obtain the estimation to signal is sent based on decoding result.
In some instances, it obtains to can be the estimation for sending signal based on decoding result and utilize and used decoding
The corresponding encoding scheme of scheme recompiles decoding result, will recompile encoded signal obtained and be used as to hair
The estimation for the number of delivering letters.In this case, sending signal estimator can be using corresponding with used decoding scheme
The encoder of encoding scheme, such as encoder identical with sending side.It certainly, herein can also be any using other in this field
It can be used for estimating the technical solution of transmission signal in receiving side, the embodiment of the present disclosure is not by the specific implementation for sending signal estimation
Limitation.
Noise estimator 230 obtains the estimation to interchannel noise for the correlation using interchannel noise.
By in view of in channel because of the correlation of the noise caused by the factors such as filtering, over-sampling, multi-user interference,
In the embodiment of the present disclosure noise in actual channel can be more accurately estimated using the correlation.
In some instances, noise estimator 230 may include the first noise estimation module 232 and the second noise estimation module
234.First noise estimation module 232 can be adder/subtracter, can be used for above-mentioned to hair by subtracting from reception signal
The estimation for the number of delivering letters is estimated to obtain the first of interchannel noise.Second noise estimation module 234 can be used for utilizing interchannel noise
Correlation obtained handle the first of interchannel noise the estimation obtained and estimated the second of interchannel noise, with
As the above-mentioned estimation to interchannel noise.
In some instances, convolutional neural networks (CNN, convolutional neural network) can be used to come in fact
Now to the utilization of the correlation of interchannel noise, also that is, the second noise estimation module 234 is embodied as convolutional neural networks.Herein
In the case of, it can be by the input to the first of interchannel noise the estimation as convolutional neural networks, with the output work of convolutional neural networks
For the second estimation to interchannel noise.
In some instances, available before being obtained using convolutional neural networks to the second estimation of interchannel noise
Predetermined policy is trained convolutional neural networks, so that convolutional neural networks can simulate the scene of realistic channels.Herein
In the case of, equipment shown in Fig. 2 may also include for using predetermined policy come the network training device of training convolutional neural networks
250.In some instances, which may include the Training strategy for minimizing surplus noise power, such as baseline BP-CNN
(Baseline BP-CNN) strategy.In other examples, predetermined policy, which also is included in, minimizes the same of surplus noise power
When make the distribution of surplus noise close to the Training strategy of Gaussian Profile, for example, enhancing BP-CNN (Enhanced BP-CNN) plan
Slightly.Certainly, the strategy that can be used for training convolutional neural networks in the embodiment of the present disclosure is not limited to both the above, but may include energy
Enough enable training after convolutional neural networks more accurately reflect reality channel any other strategy.
Since noise estimator 230 obtains the good estimation to interchannel noise using the correlation of interchannel noise, pass through
The good estimation is subtracted in receiving signal, the reception signal of modification obtained can have better access to the transmission at sending side
Send signal.
After noise estimator 230 is obtained using the correlation of interchannel noise to the estimation of interchannel noise, modification letter
Number generator 230 is used to subtract the estimation obtained to interchannel noise from receiving in signal, to obtain the reception signal of modification.
In some cases, the once-through operation of noise estimator 230 may not be obtained to actual channel noise (example
Such as, coloured noise) good estimation.In this case, it is possible to carry out iteration using the reception signal of above-mentioned modification as signal is received
It executes decoder 210, send signal estimator 220, noise estimator 230 and the operation for modifying signal generator 240, until being
System state meets pre-provisioning request.The number that pre-provisioning request described herein can be system iteration reach pre-set number or
Decoded signal no longer changes, or is also possible to other any requirements that those skilled in the art are commonly used in noise estimation.
When obtaining the reception signal of modification, decoder 210 is also using the predetermined decoding scheme to the reception signal of modification
It is decoded, to obtain decoded signal.
In equipment shown in Fig. 2, the good estimation to interchannel noise is obtained by the correlation using interchannel noise
And the good estimation is subtracted in receiving signal, the estimation to transmission signal can be relatively accurately obtained, and realize in turn
The decoding of high quality.
The technical solution of the embodiment of the present disclosure is retouched above by method shown in FIG. 1 and equipment shown in Fig. 2
It states.The technical solution according to the embodiment of the present disclosure will be described in detail by specific example below.It should be noted that this public affairs
The technical solution for opening embodiment is not limited to the specific example, but also may include that the present invention that falls in made to the example protects model
Enclose interior various modifications.
Fig. 3 shows the simplified diagram of the decoding scheme according to the embodiment of the present disclosure.It should be noted that Fig. 3 is only
For illustrating a specific example of the embodiment of the present disclosure, it is not construed as the limitation to the embodiment of the present disclosure.For example, Fig. 3 exists
Sending side has used LDPC coding and binary phase shift keying (BPSK) modulation scheme, and has used the decoding side BP in receiving side
Case, however it will be appreciated by those skilled in the art that the technical solution of the embodiment of the present disclosure can also be applied to different encoding and decoding and
Modulation system.
In the technical solution of Fig. 3, it is assumed that code word u is obtained by LDPC encoder in transmitting terminal block of information bits x, logical
BPSK is crossed to modulate to obtain symbol s.Symbol s is after coloured noise channel, and receiving end will receive noise cancellation signal y=s+n, wherein n
Indicate coloured interchannel noise.Receiving end passes through the iteration structure pair being mainly made of BP decoder and convolutional neural networks (CNN)
Information is sent to be decoded.An iteration process mainly includes two steps.The first step has noise cancellation signal y to first pass through a standard
BP decoder is denoted as according to the available estimation to symbol s is sent of decoding resultThe estimated value is received from receiving end
Have in noise cancellation signal y and subtracts, the available estimation to interchannel noise, i.e.,Because possible BP decoding error is deposited
,There are errors between real channel noise.Therefore noise estimation value can be write asξ indicates to solve due to BP
Noise estimation error caused by code mistake.Second step, willThe convolutional neural networks (CNN) of a depth are inputted, CNN will be utilized
The correlation of interchannel noise again estimates noise, obtains being denoted as the more accurate estimated value of noiseIt willFrom connecing
It is subtracted in collection of letters y availableWhereinIt is defined as surplus to make an uproar
Sound.When the comparison that CNN estimates interchannel noise is accurate, the power of surplus noise is lower,Possess signal-to-noise ratio more higher than y.Cause
This, willIt is again inputted into BP decoder, BP decoder will obtain more accurate decoding result.It can be into the above process
Row iteration, the influence of compression surplus noise gradually improve decoding performance.
In the example of fig. 3, the beginning and end of above-mentioned iterative process is shown by way of single-pole double-throw switch (SPDT).Example
Such as, when in the above-mentioned first step pass through standard BP decoder to can make when thering is noise cancellation signal y to be decoded switch connection BP, CNN and
The circulation that adder is formed to start to estimate the iterative operation of noise, and meets pre-provisioning request (for example, system in system mode
The number of iteration reaches pre-set number or decoded signal no longer changes) after can disconnect the circulation and connect have noise cancellation signal
Y and BP decoder, to terminate the iterative operation.The control signal of switch can be currently used by this field or be used in the future
Any mode generate, the limitation of the specific producing method of the uncontrolled signal of the scope of the present invention.In addition, the disclosure is implemented
The beginning and end of iterative cycles is not limited to the form of single-pole double-throw switch (SPDT) shown in Fig. 3 in example, but this field can be used
Any specific implementation form for circulate operation.
The schematic diagram of the CNN structure according to used in the embodiment of the present disclosure is shown in Fig. 4.In structure shown in Fig. 4,
Input is the vector of N × 1, i.e. noise is estimatedIn first layer, k is obtained by convolution algorithm1A characteristic pattern, mathematical form
ForWherein c1, jIndicate j-th of characteristic pattern of first layer, h1, jIt is f for length1One-dimensional volume
Product core, * indicate convolution algorithm, b1, jIndicate the corresponding departure of j-th of characteristic pattern, ReLU (Rectified Linear Unit
Function) activation primitive (i.e. max (x, 0)) is indicated.At i-th layer, convolution algorithm will be on all characteristic patterns that upper layer exports
It carries out, therefore is considered as a two-dimensional convolution, mathematical form cI, j=ReLU (hI, j*ci-1+bI, j)。cI, jIt is i-th layer
J-th of characteristic pattern, hI, jFor i-th layer of j-th of convolution kernel, size fi×ki-1, fiAnd ki-1Respectively indicate i-th layer of convolution kernel
The number c of size and (i-1)-th layer of characteristic patterni-1Indicate two-dimensional matrix made of (i-1)-th layer of all characteristic pattern arrangement.Use L table
The number of plies for showing network is in the final output of the last layer i.e. L layers, networkEstimate with noise
Estimate compared to more accurate noise.Network structure shown in Fig. 4 can be abbreviated as { L；f1, f2..., fL；k1, k2..., kL}。
As described above, can be trained to CNN network before CNN network to be used for noise estimation.The training of network can
Including two big steps.Firstly, it is necessary to generate training data for specific channel.In this step, source bits x can be with
Machine generates, and interchannel noise can acquire in actual channel, previously known channel model also can be used and be trained.In reality
In the application of border, it usually needs be trained for certain common communication scenes (model), the network model after training can be protected
There are receiving ends, to select corresponding channel model for specific communication scenes.It in actual use can be according to channel
Estimation selects suitable model to use.In the case where there are interchannel noise data, CNN can be obtained according to process shown in Fig. 3
Input dataPrimary training only can be done to CNN, use the same network model always in an iterative process, certainly,
In some examples, can also repeatedly it be trained.
In order to train network, need to define loss function.The embodiment of the present disclosure provides two kinds and defines loss function below
Method respectively corresponds different network training strategies.It should be noted that following loss function definition and/or network training
Strategy is only the example that provides, the definition of other loss functions used in the art to illustrate the scheme of the embodiment of the present disclosure
And/or network training strategy can also be applied to the embodiment of the present disclosure.
Network training strategy 1: baseline BP-CNN (Baseline BP-CNN), the strategy minimize the power of surplus noise,
Loss function at this time may be defined as
Wherein r indicates that surplus noise vector, N indicate the length of the vector.
In the Training strategy, the experience distribution of surplus noise is counted after training, and initial using this distribution
Change log-likelihood ratio (LLR, log-likelihood ratios) value of the decoded variable nodes of BP next time.
Network training strategy 2: enhancing BP-CNN (Enhanced BP-CNN), the strategy while compressing surplus noise,
The distribution for adjusting surplus noise, follows it close to Gaussian Profile.Due to current most encoders both for Gaussian channel into
Row design, which can preferably be adapted with encoder.Loss function is defined as at this time
Wherein the definition of first part is with network training strategy 1, for measuring the power of surplus noise.Wherein second part
It is tested from Jarque-Bera, a kind of common just too property method of inspection, specifically,
riIt is i-th of element of surplus noise vector r,For its mean value.λ is the weight for adjusting two-part weight
The factor.The experience distribution for not needing statistics surplus noise at this time need to only count its variance and calculate BP next time by Gaussian Profile
The initial value of the LLR of decoded variable node.
In order to further illustrate the effect of the technical solution of the embodiment of the present disclosure, Fig. 5 to Fig. 8 is respectively illustrated for two
The performance comparison schematic diagram of different channels correlation models.Wherein, Fig. 5 to Fig. 6 is shown for the first channel relevancy mould
Performance comparison schematic diagram of the type under different degrees of correlation, Fig. 7 show the performance for the first channel relevancy model
With the relation schematic diagram of the number of iterations, Fig. 8 shows the performance comparison schematic diagram for second of channel relevancy model.
The first channel relevancy model:
Element definition in the correlation matrix R of the model is RI, j=η|i-j|.LDPC encoder bit rate is 3/4, code length 576,
Encoder matrix is from WiMax standard.It should be noted that above-mentioned parameter is only example, the technical solution of the embodiment of the present disclosure
Application scenarios independent of specific encoder bit rate and encoder matrix.Under the above parameters, used CNN network structure can
To be { 4；9,3,3,15；64,32,16,1 }.It is set forth in fig. 5 and fig. (strong in parameter η=0.8 for indicating correlation
Correlation) and η=0.5 (moderate correlation) under only carry out the test result of a BP-CNN iteration.Decoding structure at this time
It can be abbreviated as BP (x)-CNN-BP (x), the number of iterations of the digital representation BP decoder in bracket.It is given in Fig. 5 and Fig. 6
BP (5)-CNN-BP (5) structure complexity out is roughly equivalent to 12 standard BP iteration, as can be seen from Figure, regardless of
It is that can be obtained using baseline BP-CNN Training strategy or enhancing BP-CNN Training strategy, the technical solution of the embodiment of the present disclosure
Result more better than standard BP algorithm.Meanwhile the number of iterations for increasing standard BP can further increase performance, but raising extremely has
Limit.This illustrates that the technical solution of the embodiment of the present disclosure can obtain higher performance with lower complexity.Fig. 7 gives progress
Multiple BP-CNN iteration as a result, as seen from the figure, multiple BP-CNN iteration can further promote decoding performance.
Second of channel relevancy model:
Power spectral density can be used to describe in the correlation of the channel correlation model, i.e. P (f) ∝ 1/ | f |α.Particularly,
As α=1, this noise is known as pink colour noise.Fig. 8 gives identical with the first channel relevancy model in remaining condition
In the case of, the technical solution of the embodiment of the present disclosure and the performance of standard BP decoder compare.As can be seen from Figure 8, the disclosure
The technical solution of embodiment remains to obtain better decoding performance under this channel model.
Fig. 9 diagrammatically illustrates the block diagram of equipment according to an embodiment of the present disclosure.Equipment shown in Fig. 9 is only one
Example, should not function to the embodiment of the present disclosure and use scope bring any restrictions.
As shown in figure 9, equipment 900 includes central processing unit (CPU) 901 according to this embodiment, it can be according to depositing
Storage is loaded into random access storage device (RAM) 903 in the program in read-only memory (ROM) 902 or from storage section 908
Program and execute various movements appropriate and processing.In RAM 903, also it is stored with equipment 900 and operates required various journeys
Sequence and data.CPU 901, ROM 902 and RAM 903 are connected with each other by bus 904.Input/output (I/O) interface 905
It is also connected to bus 904.
Equipment 900 can also include be connected to I/O interface 905 with one or more in lower component: including keyboard or
The importation 906 of mouse etc.；Including cathode-ray tube (CRT) or liquid crystal display (LCD) etc. and loudspeaker etc.
Output par, c 907；Storage section 908 including hard disk etc.；And the network interface including LAN card or modem etc.
The communications portion 909 of card.Communications portion 909 executes communication process via the network of such as internet.Driver 910 is also according to need
It is connected to I/O interface 905.Detachable media 911, such as disk, CD, magneto-optic disk or semiconductor memory etc., according to
It needs to be mounted on driver 910, in order to be mounted into storage section as needed from the computer program read thereon
908。
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be carried on computer-readable medium
On computer program, which includes the program code for method shown in execution flow chart.In such reality
It applies in example, which can be downloaded and installed from network by communications portion 909, and/or from detachable media
911 are mounted.When the computer program is executed by central processing unit (CPU) 901, execute in the equipment of the embodiment of the present disclosure
The above-mentioned function of limiting.
It should be noted that computer-readable medium shown in the disclosure can be computer-readable signal media or meter
Calculation machine readable storage medium storing program for executing either the two any combination.Computer readable storage medium for example can be --- but not
Be limited to --- electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor system, device or device, or any above combination.It calculates
The more specific example of machine readable storage medium storing program for executing can include but is not limited to: have the electrical connection, portable of one or more conducting wires
Formula computer disk, hard disk, random access storage device (RAM), read-only memory (ROM), erasable programmable read only memory
(EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device or
Above-mentioned any appropriate combination.In the disclosure, computer readable storage medium can be it is any include or storage program
Tangible medium, the program can be commanded execution system, device or device use or in connection.And in the disclosure
In, computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal, wherein
Carry computer-readable program code.The data-signal of this propagation can take various forms, including but not limited to electric
Magnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be computer-readable and deposit
Any computer-readable medium other than storage media, the computer-readable medium can send, propagate or transmit for by referring to
Enable execution system, device or device use or program in connection.The program for including on computer-readable medium
Code can transmit with any suitable medium, including but not limited to: wireless, electric wire, optical cable or RF etc. or above-mentioned times
The suitable combination of meaning.
Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the disclosure, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, section or code of table, a part of above-mentioned module, section or code include one or more use
The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box
The function of note can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are actually
It can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it to infuse
It anticipates, the combination of each box in block diagram or flow chart and the box in block diagram or flow chart can be used and execute regulation
The dedicated hardware based systems of functions or operations realize, or can be using a combination of dedicated hardware and computer instructions
To realize.
Such as field programmable gate can also be used according to the method, apparatus of each embodiment of the disclosure, unit and/or module
Array (FPGA), programmable logic array (PLA), system on chip, the system on substrate, the system in encapsulation, dedicated integrated electricity
Road (ASIC) can be for carrying out the hardware such as any other rational method that is integrated or encapsulating or firmware in fact to circuit
It is existing, or realized with software, the appropriately combined of three kinds of implementations of hardware and firmware.The system may include storage equipment,
To realize storage as described above.When realizing in such ways, used software, hardware and/or firmware be programmed or
It is designed as executing the corresponding above method, step and/or the function according to the disclosure.Those skilled in the art can be according to practical need
Come suitably by one or more of these systems and module, or in which a part or multiple portions use it is different upper
Implementation is stated to realize.These implementations each fall within protection scope of the present invention.
As the skilled person will appreciate, for any and all purpose, such as written theory is being provided
The aspect of bright book, all ranges disclosed herein are also covered by any and all possible subrange and its son
The combination of range.Any listed range, which can be readily identified into, adequately describes and enables same range
At least it is broken down into same two parts, three parts, four parts, five parts, ten parts, etc..As a non-limiting example,
Each range discussed herein can be easily decomposed into lower one third, middle one third and upper three/
One etc..Such as those skilled in the art it will also be understood that, all languages of " until ", " at least ", " being greater than ", " being less than " etc.
Yan Jun includes stated quantity and is the range for referring to be broken down into subrange as discussed above therewith.Finally,
As the skilled person will appreciate, range includes each individual ingredient.So for example, the group with 1-3 unit
Refer to the group with 1,2 or 3 unit.Similarly, the group with 1-5 unit refers to 1,2,3,4 or 5 unit
Group, etc..
Although the present invention, art technology has shown and described referring to the certain exemplary embodiments of the disclosure
Personnel it should be understood that in the case where the spirit and scope of the present invention limited without departing substantially from the following claims and their equivalents,
A variety of changes in form and details can be carried out to the disclosure.Therefore, the scope of the present invention should not necessarily be limited by above-described embodiment,
But should be not only determined by appended claims, also it is defined by the equivalent of appended claims.
Claims (12)
1. a kind of method for decoded signal, comprising:
It is decoded using predetermined decoding scheme to signal is received, and the estimation to signal is sent is obtained based on decoding result；
The estimation to interchannel noise is obtained using the correlation of interchannel noise；
The estimation obtained to interchannel noise is subtracted from the reception signal, to obtain the reception signal of modification；And
It is decoded using reception signal of the predetermined decoding scheme to the modification, to obtain decoded signal.
2. according to the method described in claim 1, wherein, the estimation to interchannel noise is obtained using the correlation of interchannel noise
Include:
The first estimation to interchannel noise is obtained by subtracting the estimation of described pair of transmission signal from the reception signal；With
And
The first estimation obtained to interchannel noise is handled using the correlation of interchannel noise to obtain to channel
Second estimation of noise, using as the estimation to interchannel noise.
3. according to the method described in claim 2, wherein, using the correlation of interchannel noise come to obtained to interchannel noise
The first estimation handle obtain estimate the second of interchannel noise include:
By the input to the first estimation of interchannel noise as convolutional neural networks, with the output of the convolutional neural networks
As second estimation to interchannel noise.
4. according to the method described in claim 3, further include:
The convolutional neural networks are trained using predetermined policy, wherein the predetermined policy includes any one of following: being minimized
The Training strategy of surplus noise power, and make the distribution of surplus noise close to Gauss while minimizing surplus noise power
The Training strategy of distribution.
5. according to the method described in claim 1, further include:
The reception signal of the modification is carried out into iteration as reception signal and executes the decoding of reception signal, based on the phase of interchannel noise
Acquisition of the closing property to the reception signal of estimation and the modification of interchannel noise, until system mode meets pre-provisioning request.
6. a kind of equipment for decoded signal, comprising:
Decoder, for being decoded using predetermined decoding scheme to signal is received；
Signal estimator is sent, for obtaining the estimation to signal is sent based on decoding result；
Noise estimator obtains the estimation to interchannel noise for the correlation using interchannel noise；And
Signal generator is modified, for subtracting the estimation obtained to interchannel noise from the reception signal, to be repaired
The reception signal changed；
Wherein, the decoder is also decoded using reception signal of the predetermined decoding scheme to the modification, to obtain
Decoded signal.
7. equipment according to claim 6, wherein the noise estimator includes:
First noise estimation module, for being obtained pair by subtracting the estimation of described pair of transmission signal from the reception signal
First estimation of interchannel noise；And
Second noise estimation module, for the correlation using interchannel noise come to the first estimation obtained to interchannel noise
It is handled to obtain to the second of interchannel noise the estimation, using as the estimation to interchannel noise.
8. equipment according to claim 7, wherein the second noise estimation module is realized using convolutional neural networks
, the second noise estimation module is also used to:
By the input to the first estimation of interchannel noise as the convolutional neural networks, with the convolutional neural networks
It exports as second estimation to interchannel noise.
9. equipment according to claim 8, further includes:
Network training device, for training the convolutional neural networks using predetermined policy, wherein the predetermined policy include with
It is any one of lower: to minimize the Training strategy of surplus noise power, and so that surplus is made an uproar while minimizing surplus noise power
Training strategy of the distribution of sound close to Gaussian Profile.
10. equipment according to claim 6, wherein using the reception signal of the modification as signal is received, iteratively hold
The row decoder, the operation for sending signal estimator, the noise estimator and the modification signal generator, until
System mode meets pre-provisioning request.
11. a kind of equipment for decoded signal, comprising:
Memory, for storing executable instruction；And
Processor, for executing the executable instruction stored in memory, to execute according to claim 1 described in any one of -5
Method.
12. one kind carries the memory devices of computer program, when executing the computer program by processor, institute thereon
Stating computer program makes the processor execute method according to any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710753622.5A CN109428673A (en) | 2017-08-28 | 2017-08-28 | For the method for decoded signal, equipment and storage equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710753622.5A CN109428673A (en) | 2017-08-28 | 2017-08-28 | For the method for decoded signal, equipment and storage equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109428673A true CN109428673A (en) | 2019-03-05 |
Family
ID=65501886
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710753622.5A Pending CN109428673A (en) | 2017-08-28 | 2017-08-28 | For the method for decoded signal, equipment and storage equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109428673A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110445581A (en) * | 2019-08-10 | 2019-11-12 | 南京理工大学 | The method for reducing the channel decoding bit error rate based on convolutional neural networks |
| CN110752894A (en) * | 2019-10-17 | 2020-02-04 | 网络通信与安全紫金山实验室 | CNN-based LDPC code blind channel decoding method and decoder |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1391220A (en) * | 2001-06-11 | 2003-01-15 | 富士通株式会社 | Information record and reproduction device, method and signal decoding circuit |
| CN101552939A (en) * | 2009-05-13 | 2009-10-07 | 吉林大学 | In-vehicle sound quality self-adapting active control system and method |
| US7738537B2 (en) * | 2005-09-14 | 2010-06-15 | Novatel Inc. | Apparatus for and method of determining quadrature code timing from pulse-shape measurements made using an in-phase code |
| CN102461037A (en) * | 2009-06-16 | 2012-05-16 | 夏普株式会社 | Wireless communication apparatus, wireless communication system and communication method |
| CN103297361A (en) * | 2013-02-19 | 2013-09-11 | 上海晨思电子科技有限公司 | Channel and noise estimation method and channel and noise estimation device |
| CN104182948A (en) * | 2013-12-23 | 2014-12-03 | 上海联影医疗科技有限公司 | Estimation method of correlation noise |
| CN106571831A (en) * | 2016-10-28 | 2017-04-19 | 华南理工大学 | LDPC hard decision decoding method based on depth learning and decoder |
-
2017
- 2017-08-28 CN CN201710753622.5A patent/CN109428673A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1391220A (en) * | 2001-06-11 | 2003-01-15 | 富士通株式会社 | Information record and reproduction device, method and signal decoding circuit |
| US7738537B2 (en) * | 2005-09-14 | 2010-06-15 | Novatel Inc. | Apparatus for and method of determining quadrature code timing from pulse-shape measurements made using an in-phase code |
| CN101552939A (en) * | 2009-05-13 | 2009-10-07 | 吉林大学 | In-vehicle sound quality self-adapting active control system and method |
| CN102461037A (en) * | 2009-06-16 | 2012-05-16 | 夏普株式会社 | Wireless communication apparatus, wireless communication system and communication method |
| CN103297361A (en) * | 2013-02-19 | 2013-09-11 | 上海晨思电子科技有限公司 | Channel and noise estimation method and channel and noise estimation device |
| CN104182948A (en) * | 2013-12-23 | 2014-12-03 | 上海联影医疗科技有限公司 | Estimation method of correlation noise |
| CN106571831A (en) * | 2016-10-28 | 2017-04-19 | 华南理工大学 | LDPC hard decision decoding method based on depth learning and decoder |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110445581A (en) * | 2019-08-10 | 2019-11-12 | 南京理工大学 | The method for reducing the channel decoding bit error rate based on convolutional neural networks |
| CN110445581B (en) * | 2019-08-10 | 2022-11-01 | 南京理工大学 | Method for reducing channel decoding error rate based on convolutional neural network |
| CN110752894A (en) * | 2019-10-17 | 2020-02-04 | 网络通信与安全紫金山实验室 | CNN-based LDPC code blind channel decoding method and decoder |
| CN110752894B (en) * | 2019-10-17 | 2022-04-29 | 网络通信与安全紫金山实验室 | CNN-based LDPC code blind channel decoding method and decoder |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107632987B (en) | A kind of dialogue generation method and device | |
| US20180241209A1 (en) | Dispatching method and device for integrated transmission and distribution network | |
| US20180357530A1 (en) | Deep learning decoding of error correcting codes | |
| CN108090521B (en) | Image fusion method and discriminator of generative confrontation network model | |
| US10924304B2 (en) | Method and device for decoding a signal, and memory device | |
| CN113473149A (en) | Semantic channel joint coding method and device for wireless image transmission | |
| CN109361404A (en) | A kind of LDPC decoding system and interpretation method based on semi-supervised deep learning network | |
| CN111224677B (en) | Encoding method, decoding method and device | |
| CN109428673A (en) | For the method for decoded signal, equipment and storage equipment | |
| CN107078747A (en) | The streamline layering LDPC decodings arbitrated with pre-configured memory | |
| Mashhadi et al. | Level crossing speech sampling and its sparsity promoting reconstruction using an iterative method with adaptive thresholding | |
| US11336300B2 (en) | Generalized polar codes | |
| WO2023279790A1 (en) | Method for transmitting transformer process design file | |
| CN108631790A (en) | A kind of LDPC code building method and device | |
| CN104348764B (en) | The method and apparatus that computing unit is distributed in data receiver link | |
| CN110798857A (en) | Simulation method for influence of 5G slice network on service | |
| CN108055045A (en) | A kind of structure of new ldpc code decoder | |
| CN110034843B (en) | Channel coding method and coding device | |
| CN107633219A (en) | Integrated optical character identifying method and system | |
| CN109547156A (en) | Non-coherent detection methods, device and the electronic equipment of polar short code | |
| CN107018287A (en) | The method and apparatus for carrying out noise reduction to image using video epitome | |
| US10559093B2 (en) | Selecting encoding options | |
| Dai et al. | Optimisation design of systematic fountain codes on fading channels | |
| CN110033413A (en) | Image processing method, device, electronic equipment, the computer-readable medium of client | |
| Uçan et al. | Turbo coded signals over wireless local loop environment |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190305 |
|
| WD01 | Invention patent application deemed withdrawn after publication |