CN115955279B - Channel matching non-gridding coding modulation method based on variable self-encoder - Google Patents
Channel matching non-gridding coding modulation method based on variable self-encoder Download PDFInfo
- Publication number
- CN115955279B CN115955279B CN202310242421.4A CN202310242421A CN115955279B CN 115955279 B CN115955279 B CN 115955279B CN 202310242421 A CN202310242421 A CN 202310242421A CN 115955279 B CN115955279 B CN 115955279B
- Authority
- CN
- China
- Prior art keywords
- distribution
- encoder
- probability
- channel
- variable
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000009826 distribution Methods 0.000 claims abstract description 59
- 239000000835 fiber Substances 0.000 claims abstract description 35
- 230000005540 biological transmission Effects 0.000 claims abstract description 22
- 230000008569 process Effects 0.000 claims abstract description 12
- 238000004088 simulation Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 14
- 238000005457 optimization Methods 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 5
- 230000001172 regenerating effect Effects 0.000 claims description 2
- 238000007493 shaping process Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 239000013307 optical fiber Substances 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 210000000709 aorta Anatomy 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000009022 nonlinear effect Effects 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 230000003313 weakening effect Effects 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Digital Transmission Methods That Use Modulated Carrier Waves (AREA)
- Optical Communication System (AREA)
Abstract
The invention discloses a channel matching non-gridding coding modulation method based on a variable self-encoder, which comprises the following steps: acquiring a non-meshed signal, inputting the non-meshed signal into a pre-trained variable self-encoder, performing geometric probability encoding based on the geometric position and the distribution probability of a constellation of the non-meshed signal, generating a multi-dimensional parameter, and inputting the multi-dimensional parameter into a preset nonlinear simulation fiber channel; after nonlinear simulation fiber channel, the fiber channel enters an isomorphic variable self-decoder, and generalized mutual information of a transmission link is calculated; and (3) performing generalized mutual information judgment, and outputting a current signal if the current signal constellation distribution is the optimal non-meshed constellation distribution of channel matching or the iteration number of multidimensional parameters reaches a preset value. The invention realizes the mixed forming scheme of probability forming and geometric forming by means of the variable self-encoder, and the optimal forming scheme of channel matching is iterated out intelligently according to channel feedback, so as to realize the maximization of the generalized mutual information of the link, and finally, the non-gridding signal coding modulation process of channel matching is completed.
Description
Technical Field
The invention relates to a channel matching non-gridding coding modulation method based on a variable self-encoder, belonging to the technical field of hybrid forming technology and machine learning.
Background
The optical fiber communication system is a bearing foundation stone for information transmission and is an aorta of a broadband network. In recent years, new services such as metauniverse, digital twin, industrial internet and the like are continuously developed, network traffic is increased explosively, and according to statistics, the global network traffic reaches 150.7TB per second in the future. The communication system is used as a physical foundation for bearing the global network, the 5G and later 5G wireless communication technologies are rapidly developed, and the capacity of an access side is improved by more than 1000 times compared with that of 4G. Because the data of the access side is finally collected to the backbone network, more than 85% of the data of the backbone network in the world is transmitted in long distance through the optical fiber, and therefore, the capacity expansion of the optical fiber communication system is imperative. The current optical transmission system has realized the upgrade from 100Gb/s to 200Gb/s of single channel, and the current 400Gb/s and 800Gb/s of single channel have been initially commercialized. To further increase the capacity, the current common method is to increase the modulation order of the signal, and the fiber channel is used as a nonlinear channel, and nonlinear phase noise caused by self-phase modulation and cross-phase modulation severely restricts long-distance transmission of the high-order modulated signal. Although a large number of methods such as probability shaping, geometric shaping, distributed Fourier expansion and the like are proposed by research groups at home and abroad to further approach the nonlinear shannon limit of the optical fiber, the technologies are solidified in an optimized mode of a transmitting end modulation format, and intelligent matching of high-order modulation and nonlinear optical fiber channels is not considered, so that high-gain long-distance transmission of high-order modulation signals is difficult to realize.
Non-meshed constellation design is an effective means of approaching shannon's limit. Non-meshed constellation hybrid shaping includes probabilistic shaping and geometric shaping. Probability shaping is carried out by adjusting the prior probability of constellation points, so that the transmission probability is subjected to Gaussian distribution, namely, low-energy points are transmitted with high probability, and high-energy points are transmitted with low probability, thereby not only effectively improving the error code performance of a system, but also remarkably saving the transmission power, and weakening the influence of nonlinear effects on signals to a certain extent. The geometric shaping is to adjust the constellation point position to make the performance superior to the constellation point geometric position distribution of the traditional constellation diagram, reduce the required transmitting power and obtain the generalized mutual information gain by considering the factors of the minimum Euclidean distance, the average energy of the constellation, the channel noise characteristic, the coding mapping, the symmetry and the like. Both probability shaping and geometric shaping are effective new ways for improving the system capacity and the system performance. However, the gain achieved by the related studies is not sufficient to reach the limit, and the gap from the theoretical limit capacity cannot be completely filled. From the aspect of action mechanism, the traditional constellation forming is to directly modulate the constellation into a channel after the constellation is optimized in advance, the actual channel is in a change, the optimized constellation which is not matched with the corresponding condition in time is in a non-applicable environment, and the performance in actual application is greatly reduced; from the aspect of implementation method, the position and probability of the direct design constellation cannot meet the requirements, and the dynamic influence of the channel on constellation optimization cannot be handled by a small amount of researches on improving constellation parameters by using an optimization algorithm. The high-order modulation format of the traditional high-order modulation design is limited to the consideration of the modulation format of the transmitting end, and is solidified in an optimized mode and cannot be matched with the channel condition. Moreover, the traditional constellation shaping technology is difficult to combine geometric shaping and probability shaping technologies, a plurality of influencing parameters such as distribution probability, geometric position, euclidean distance, channel response, transmission distance and the like are isolated from each other, and intelligent matching of high-order modulation and nonlinear fiber channels is not considered.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, and provides a channel matching non-gridding coding modulation method based on a variable self-encoder, which performs simultaneous optimization on multidimensional influence parameters of an optical communication system such as channel response, distribution probability, geometric position and the like, realizes a mixed forming scheme of probability forming and geometric forming by means of the variable self-encoder, and iterates out an optimal forming scheme of channel matching according to channel feedback intelligence, realizes maximization of link generalized mutual information and finally completes a non-gridding signal coding modulation process of channel matching.
In order to achieve the above object, the present invention provides a channel matching non-trellis coded modulation method based on a variable self-encoder, comprising the steps of:
acquiring a non-meshed signal, inputting the non-meshed signal into a pre-trained variable self-encoder, performing geometric probability encoding based on the geometric position and the distribution probability of a constellation of the non-meshed signal, generating a multi-dimensional parameter, and inputting the multi-dimensional parameter into a preset nonlinear simulation fiber channel;
after nonlinear simulation fiber channel, the fiber channel enters an isomorphic variable self-decoder, geometric probability decoding is carried out on multidimensional parameters, the multidimensional parameters are output, and generalized mutual information of a transmission link is calculated;
performing generalized mutual information judgment, and outputting a current signal if the current signal constellation distribution is the optimal non-meshed constellation distribution of channel matching or the iteration number of multidimensional parameters reaches a preset value;
otherwise, repeating the steps after updating the iterative multidimensional parameter based on the generalized mutual information.
Further, the pre-training step of the variable self-encoder includes:
model training is carried out by adopting an end-to-end network architecture, and a mean value vector and a standard deviation vector are obtained in the encoding process by utilizing parameter reconstruction;
sampling from the standard deviation vector, and adding the sampling result to the mean vector to realize the coding process of the potential variable.
Further, the geometric probability coding is performed based on the geometric position and the distribution probability of the constellation, and the multi-dimensional parameter is generated, which comprises the following steps:
marking constellation distribution probability and geometric position of non-meshed signals, and sending the constellation distribution probability and geometric position as input data to a variable self-encoder;
the variable self-encoder converts the input data into a probability distribution of the latent variables, then samples the probability distribution as the latent variables, and converts the latent variables into reconstructed output.
Further, the variable self-encoder converts the input data into a probability distribution of latent variables, comprising:
obtaining a selection distribution of the potential variable, and performing a similarity operation on the probability distribution and the selection distribution of the potential variable by utilizing KL divergence in a loss function of the variable self-encoder to obtain the probability distribution of the potential variable:
wherein ,
reconstruction loss, i.e. input data X and reconstruction output +.>
Mean square error between;
for probability distribution P (Z) and selection distribution +.>
KL divergence of (2); j is the number of times of inputting data; z is a potential variable of the input data X.
Further, the preset nonlinear simulated fiber channel comprises:
simulating the environment of a nonlinear fiber channel based on generating a simulated fiber channel against the preset nonlinearity of the network;
the generation countermeasure network comprises a generator and a discriminator;
training the generation countermeasure network until the structure of the generator meets the requirement of the nonlinear fiber channel, namely, the generator is used as the nonlinear fiber channel.
Further, the input of the generator includes random noise and a condition tag, and the input of the arbiter includes real data and the output of the generator.
Further, the training to generate the countermeasure network includes:
the generator generates random noise into generation data conforming to the conditions according to the feedback of the condition tag and the discriminator, and the discriminator distinguishes whether the input data is real data or generation data according to the distribution probability of the real data and feeds the data back to the generator.
Further, the calculating generalized mutual information of the transmission link includes:
wherein H (X) is the entropy of the input data X;
a kth transmission symbol that is an ith bit; />
Is given the received symbol y k Estimated probability of->
The method comprises the steps of carrying out a first treatment on the surface of the m is the number of bits per symbol; k is the number of transmission symbols;
further, the updating the iterative multidimensional parameter based on the generalized mutual information includes:
and updating parameters of the input signals based on the generalized mutual information, performing iterative optimization, generating updated geometric positions and distribution probabilities, and regenerating multidimensional parameters.
Further, the optimal non-meshed constellation distribution of the channel matching is:
maximizing link generalized mutual information is the best non-meshed signal constellation morphology matching the nonlinear fibre channel.
The invention has the beneficial effects that:
the invention provides a channel matching non-gridding coding modulation method based on a variable self-encoder, which carries out multidimensional parameter combination on geometric positions and distribution probability, intelligently designs constellation forms of non-gridding mixed constellation forming signals, and utilizes the variable self-encoder in machine learning to complete mixed forming constellation design based on geometric forming and probability forming. And according to feedback of the nonlinear fiber channel, iterating out geometric probability distribution parameters conforming to channel characteristics, taking maximized link generalized mutual information as a target, finally completing non-gridding coding modulation of channel matching, remarkably enhancing constellation gain of long-distance transmission of a high-order modulation format, improving capacity-distance product of non-gridding mixed constellation forming signals, and further approaching Shannon limit.
Drawings
Fig. 1 is a schematic structural diagram of a channel matching non-trellis-based coded modulation method based on a variable self-encoder according to an embodiment of the present invention;
fig. 2 is a schematic diagram of transceiver encoding and decoding in a variable self-encoder-based channel matching non-trellis coded modulation method according to an embodiment of the present invention;
FIG. 3 is a flowchart of iterative update of multidimensional parameters in a variable self-encoder based channel matching non-meshing coded modulation method provided by an embodiment of the present invention;
FIG. 4 is a schematic diagram of generating an countermeasure network in a variable self-encoder based channel matching non-trellis coded modulation method provided by an embodiment of the present invention;
fig. 5 is a schematic diagram of a variable self-encoder based channel matching non-trellis coded modulation method using a generator as a non-linear simulated fiber channel according to an embodiment of the present invention.
Description of the embodiments
The invention is further described below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.
The embodiment of the invention provides a channel matching non-meshing coding modulation method based on a variable self-encoder, which is shown in fig. 1 to 5 and comprises the following steps:
step 1: the method comprises the steps of obtaining a non-meshed signal, inputting the non-meshed signal into a pre-trained variable self-encoder, carrying out geometric probability encoding based on the geometric position and the distribution probability of a constellation of the non-meshed signal, generating multidimensional parameters, and inputting the multidimensional parameters into a preset nonlinear simulation fiber channel:
as shown in fig. 2, the present patent marks the constellation distribution probability of the non-meshed signal m-QAM as [ P1, P2,..the., pm ], the geometric position as [ C1, C2,..the., cm ], and sends the same as input data [ S1, S2,..s 2m ] to the input of the pre-trained variable self-encoder. Here, input data [ S1, S2, ], S2m ] is denoted as X, and Z represents a latent variable of the input data X. The variable auto-encoder converts the input X into a probability distribution of the latent variable P (z|x) which is then randomly sampled as the latent variable Z, which is then converted into a reconstructed output. Since it is difficult to calculate P (z|x) directly, the present invention selects an easily handled approximation distribution q (z|x) to calculate, parameters of q (z|x) and P (z|x) approximate, and makes use of KL divergence in the loss function to make the two distributions as similar as possible. The loss function of the variable self-encoder is shown in equation (1):
wherein in formula (1)
Representing reconstruction losses, i.e. input data X and reconstruction output +>
Mean square error between; />
For probability distribution P (Z) and selection distribution +.>
KL divergence of j is the number of times data is input. Equation (1) will avoid those encoded reconstructed data in potential space that do not have any meaning and that are close to a priori distribution while still preserving the implicit characteristics of the input data.
When the variable self-encoder is pre-trained, the model training is performed by adopting an end-to-end network architecture. And during training, a parameter reconstruction skill is utilized to obtain a mean value vector and a standard deviation vector in the coding process, the mean value vector and the standard deviation vector are sampled, and then the mean value vector is added to realize the coding process of potential variables. The initial learning rate is set to 0.001, and is reduced by 20% every 10 training rounds; optimizing a parameter updating process by using an Adam optimizer; the batch size is set to 64, for a total of 1000 training rounds, related to the specific graphics card storage. The channel matching scheme based on the variable self-encoder adopted by the invention can lead potential variables to have richer potential information, and improve the matching degree of the generated multidimensional parameters and the nonlinear fiber channel.
Step two: after nonlinear simulation fiber channel, the fiber channel enters an isomorphic variable self-decoder, geometric probability decoding is carried out on multidimensional parameters, the multidimensional parameters are output, and generalized mutual information of a transmission link is calculated:
the Generalized Mutual Information (GMI) of the transmission link in the present patent is calculated according to formula (2), wherein H (X) is the entropy of the input data X,
the kth transmission symbol, which is the ith bit, is +.>
Is the estimated probability of a given received symbol yk
M is the number of bits per symbol and K is the number of transmitted symbols. For the case of formula (2)
The invention provides a calculation method in the formula (3):
as shown in fig. 4 and 5, the present invention simulates the environment of a nonlinear fibre channel by generating a simulated fibre channel that opposes a preset nonlinearity of the network. The multi-dimensional parameters output by the variable self-encoder at the transmitting end are used as input data of a generator in the generating countermeasure network, the multi-dimensional variables after the generating countermeasure network are equivalent to nonlinear fiber channels with preset transmission distances, and the multi-dimensional parameters are used as input of the variable self-decoder at the receiving end. Compared with the traditional step Fourier method, the method is faster and more efficient, has low computational complexity, and is convenient for updating, iterating and optimizing multidimensional parameters. The results and training process of generating an countermeasure network are shown in fig. 4, the generating of the countermeasure network including a generator and a arbiter. The inputs to the generator include random noise and condition labels, and the inputs to the arbiter include the actual data and the output of the generator. The generator generates random noise into the generated data conforming to the condition according to the feedback of the condition label and the discriminator; the discriminator distinguishes whether the input data is real data or generated data according to the distribution probability of the real data, and feeds back to the generator. The game process is used for training continuously, so that the structure of the generator is finally enabled to accord with the characteristics of the nonlinear fiber channel, and the generator is intercepted to serve as the nonlinear fiber channel.
Step three: performing generalized mutual information judgment, and outputting a current signal if the current signal constellation distribution is the optimal non-meshed constellation distribution of channel matching or the iteration number of multidimensional parameters reaches a preset value; otherwise, repeating the steps after updating the iterative multidimensional parameter based on the generalized mutual information:
as shown in fig. 3, in the flowchart of multi-dimensional parameter iterative updating, through the steps of encoding and decoding, the reconstructed non-meshed constellation form is sent into the nonlinear optical fiber link again for transmission, updating iteration is performed to continuously optimize the multi-dimensional parameter so as to maximize the link generalized mutual information, and if the current constellation distribution is judged to be the optimal non-meshed constellation distribution matched with the channel or the optimization iteration number reaches a preset value, the current non-meshed constellation multi-dimensional parameter is output.
As shown in fig. 3, when updating the iterative multidimensional parameter, the parameters of the input signal are updated based on the generalized mutual information and are subjected to iterative optimization, so that updated geometric positions and distribution probabilities are generated, and the multidimensional parameter is regenerated.
The invention carries out iterative optimization on input parameters through the link generalized mutual information, and continuously reduces the loss function to maximize the link generalized mutual information until the optimal non-meshed signal constellation morphology of the nonlinear fiber channel is matched.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (7)
1. A channel matching non-gridding coding modulation method based on a variable self-encoder is characterized in that: the method comprises the following steps:
acquiring a non-meshed signal, inputting the non-meshed signal into a pre-trained variable self-encoder, performing geometric probability encoding based on the geometric position and the distribution probability of a constellation of the non-meshed signal, generating a multi-dimensional parameter, and inputting the multi-dimensional parameter into a preset nonlinear simulation fiber channel;
the geometric probability coding is performed based on the geometric position and the distribution probability of the constellation, and the multi-dimensional parameter generation comprises the following steps:
marking constellation distribution probability and geometric position of non-meshed signals, and sending the constellation distribution probability and geometric position as input data to a variable self-encoder;
introducing potential variables of the input data, converting the input data into probability distribution of the potential variables by the variable self-encoder, sampling the probability distribution into the potential variables, and converting the potential variables into reconstructed output;
the variable self-encoder converts input data into a probability distribution of latent variables, comprising:
obtaining a selection distribution of the potential variable, and performing a similarity operation on the probability distribution and the selection distribution of the potential variable by utilizing KL divergence in a loss function of the variable self-encoder to obtain the probability distribution of the potential variable:
wherein ,
reconstruction loss, i.e. input data X and reconstruction output +.>
Mean square error between;
for probability distribution P (Z) and selection distribution +.>
KL divergence of (2); j is the number of times of inputting data; z is a latent variable of the input data X;
after nonlinear simulation fiber channel, the fiber channel enters an isomorphic variable self-decoder, geometric probability decoding is carried out on multidimensional parameters, the multidimensional parameters are output, and generalized mutual information of a transmission link is calculated;
the calculating the generalized mutual information of the transmission link comprises the following steps:
wherein H (X) is the entropy of the input data X;
a kth transmission symbol that is an ith bit; />
Is given the received symbol y k Estimated probability of->
The method comprises the steps of carrying out a first treatment on the surface of the m is the number of bits per symbol; k is the number of transmission symbols;
performing generalized mutual information judgment, and outputting a current signal if the current signal constellation distribution is the optimal non-meshed constellation distribution of channel matching or the update iteration number of the multidimensional parameter reaches a preset value;
otherwise, repeating the steps after updating the iterative multidimensional parameter based on the generalized mutual information.
2. The variable self-encoder based channel matching non-trellis coded modulation method of claim 1, wherein:
the pre-training step of the variable self-encoder comprises:
model training is carried out by adopting an end-to-end network architecture, and a mean value vector and a standard deviation vector are obtained in the encoding process by utilizing parameter reconstruction;
sampling from the standard deviation vector, and adding the sampling result to the mean vector to realize the coding process of the potential variable.
3. The variable self-encoder based channel matching non-trellis coded modulation method of claim 1, wherein:
the preset nonlinear simulated fiber channel comprises:
simulating the environment of a nonlinear fiber channel based on generating a simulated fiber channel against the preset nonlinearity of the network; the generation countermeasure network comprises a generator and a discriminator;
training the generation countermeasure network until the structure of the generator meets the requirement of the nonlinear fiber channel, namely, the generator is used as the nonlinear fiber channel.
4. The variable self-encoder based channel matching non-trellis coded modulation method of claim 3, wherein:
the inputs to the generator include random noise and condition labels, and the inputs to the arbiter include real data and the output of the generator.
5. The variable self-encoder based channel matching non-trellis coded modulation method of claim 4, wherein:
the training of the generation of the countermeasure network comprises:
the generator generates random noise into generation data conforming to the conditions according to the feedback of the condition tag and the discriminator, and the discriminator distinguishes whether the input data is real data or generation data according to the distribution probability of the real data and feeds the data back to the generator.
6. The variable self-encoder based channel matching non-trellis coded modulation method of claim 1, wherein:
the updating of the multidimensional parameter based on the generalized mutual information comprises the following steps:
and updating parameters of the input signals based on the generalized mutual information, performing iterative optimization, generating updated geometric positions and distribution probabilities, and regenerating multidimensional parameters.
7. The variable self-encoder based channel matching non-trellis coded modulation method of claim 6, wherein:
the optimal non-meshed constellation distribution of the channel matching is as follows:
maximizing link generalized mutual information is the best non-meshed signal constellation morphology matching the nonlinear fibre channel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310242421.4A CN115955279B (en) | 2023-03-14 | 2023-03-14 | Channel matching non-gridding coding modulation method based on variable self-encoder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310242421.4A CN115955279B (en) | 2023-03-14 | 2023-03-14 | Channel matching non-gridding coding modulation method based on variable self-encoder |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115955279A CN115955279A (en) | 2023-04-11 |
| CN115955279B true CN115955279B (en) | 2023-06-02 |
Family
ID=85891422
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310242421.4A Active CN115955279B (en) | 2023-03-14 | 2023-03-14 | Channel matching non-gridding coding modulation method based on variable self-encoder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115955279B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112422195A (en) * | 2020-11-18 | 2021-02-26 | 合肥工业大学 | Design method and application of multidimensional signal modulation circuit based on generalized mutual information |
| CN112910562A (en) * | 2021-01-15 | 2021-06-04 | 清华大学深圳国际研究生院 | Communication method based on probability shaping |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109474346B (en) * | 2018-11-22 | 2020-05-22 | 北京邮电大学 | Optical signal generation method and device, electronic equipment and storage medium |
| CN111628789B (en) * | 2020-05-26 | 2023-03-28 | 复旦大学 | Signal generation method and device based on mixed probability shaping and geometric shaping |
| CN112737686B (en) * | 2021-04-01 | 2021-07-13 | 南京信息工程大学 | High-performance space optical transmission system based on geometric probability shaping technology |
| CN114499755B (en) * | 2022-01-05 | 2023-06-27 | 上海交通大学 | Geometric and probability combined constellation shaping method and system based on mutual information quantity estimation |
-
2023
- 2023-03-14 CN CN202310242421.4A patent/CN115955279B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112422195A (en) * | 2020-11-18 | 2021-02-26 | 合肥工业大学 | Design method and application of multidimensional signal modulation circuit based on generalized mutual information |
| CN112910562A (en) * | 2021-01-15 | 2021-06-04 | 清华大学深圳国际研究生院 | Communication method based on probability shaping |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115955279A (en) | 2023-04-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dai et al. | Deep learning for wireless communications: An emerging interdisciplinary paradigm | |
| Gümüş et al. | End-to-end learning of geometrical shaping maximizing generalized mutual information | |
| Mohamed et al. | Model-based: End-to-end molecular communication system through deep reinforcement learning auto encoder | |
| CN110190909B (en) | Signal equalization method and device for optical communication | |
| CN111064516B (en) | Radio over fiber communication method based on hexagonal constellation forming iteration | |
| CN110445581A (en) | The method for reducing the channel decoding bit error rate based on convolutional neural networks | |
| CN112910562A (en) | Communication method based on probability shaping | |
| CN113490223A (en) | Wireless energy supply backscattering communication method and system based on IRS | |
| CN111711455A (en) | Polarization code BP decoding method based on neural network | |
| An et al. | A learning-based end-to-end wireless communication system utilizing a deep neural network channel module | |
| CN106254030A (en) | The two-way coding and decoding method of the code of Spinal without speed | |
| CN115955279B (en) | Channel matching non-gridding coding modulation method based on variable self-encoder | |
| Song et al. | Physics-informed neural operator for fast and scalable optical fiber channel modelling in multi-span transmission | |
| CN105635006A (en) | Wavelet blind equalization method based on DNA glowworm swarm optimization algorithm | |
| CN103888392A (en) | Orthogonal wavelet transform constant modulus blind equalization algorithm based on optimization of DAN genetic algorithm | |
| CN105007246A (en) | Multi-modulus blind equalization algorithm (MMA) optimized by Memetic algorithm (MA) | |
| CN116938662A (en) | Constellation probability shaping method and device based on recurrent neural network training optimization | |
| CN105007247A (en) | Frequency domain weighted multi-modulus method for optimizing DNA sequence of novel varied DNA genetic artificial fish swarm | |
| JP7455240B2 (en) | Methods, systems, and computer programs for optimizing communication channel capacity using Dirichlet processes | |
| CN111130697B (en) | Method for reducing complexity of communication physical layer transmission system based on automatic encoder | |
| CN116128070A (en) | Federal learning method based on wireless air calculation and multi-bit quantization compressed sensing | |
| EP3926910B1 (en) | Method for optimizing the capacity of communication channels | |
| CN113766492A (en) | Safe Massive MIMO network resource allocation method | |
| CN107888537B (en) | Signal detection method for improving system complexity in large-scale antenna system | |
| Chen et al. | Integrating SLM algorithm into neural network for reducing the PAPR of OFDM System |
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 |