CN108599865B - Signal modulation format identification method and device based on photon neural network - Google Patents

Signal modulation format identification method and device based on photon neural network Download PDF

Info

Publication number
CN108599865B
CN108599865B CN201810332866.0A CN201810332866A CN108599865B CN 108599865 B CN108599865 B CN 108599865B CN 201810332866 A CN201810332866 A CN 201810332866A CN 108599865 B CN108599865 B CN 108599865B
Authority
CN
China
Prior art keywords
matrix
signal
neural network
modulation format
mach
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
Application number
CN201810332866.0A
Other languages
Chinese (zh)
Other versions
CN108599865A (en
Inventor
张天
蓝邱宇翔
戴键
李建强
周月
戴一堂
尹飞飞
徐坤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing University of Posts and Telecommunications
Original Assignee
Beijing University of Posts and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing University of Posts and Telecommunications filed Critical Beijing University of Posts and Telecommunications
Priority to CN201810332866.0A priority Critical patent/CN108599865B/en
Publication of CN108599865A publication Critical patent/CN108599865A/en
Application granted granted Critical
Publication of CN108599865B publication Critical patent/CN108599865B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
    • H04B10/60—Receivers
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
    • H04B10/60—Receivers
    • H04B10/66—Non-coherent receivers, e.g. using direct detection
    • H04B10/67—Optical arrangements in the receiver
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
    • H04B10/60—Receivers
    • H04B10/66—Non-coherent receivers, e.g. using direct detection
    • H04B10/69—Electrical arrangements in the receiver
    • H04B10/695—Arrangements for optimizing the decision element in the receiver, e.g. by using automatic threshold control
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Optical Modulation, Optical Deflection, Nonlinear Optics, Optical Demodulation, Optical Logic Elements (AREA)
  • Optical Communication System (AREA)

Abstract

The embodiment of the invention provides a signal modulation format identification method and device based on a photon neural network, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring a characteristic signal of a signal to be identified, wherein the characteristic signal is obtained after sampling the signal to be identified; inputting the characteristic signal into a photonic chip, wherein the photonic chip is used for completing linear operation in a neural network algorithm for signal modulation format identification, obtaining an output result of the photonic chip after the linear operation, and performing nonlinear operation in the neural network algorithm based on the output result to obtain a modulation format of the signal to be identified. Therefore, a processor is not needed to process matrix operation contained in a linear operation part in the neural network, the speed of signal modulation format identification is improved, and the energy consumption is low when a photonic chip is used for operation, so that the power consumption of the whole system is reduced by using the method.

Description

Signal modulation format identification method and device based on photon neural network
Technical Field
The present invention relates to the field of optical communication technologies, and in particular, to a signal modulation format recognition method and apparatus based on a photonic neural network, an electronic device, and a storage medium.
Background
With the rapid development of internet technology, the traditional wavelength division multiplexing network gradually cannot meet the service requirements, and the existing flexible optical network can effectively make up for the defects of the traditional network, thereby improving the transmission capability and the spectrum resource efficiency. Fast modulation format conversion is generally introduced into flexible optical networks, and with the rapid development of flexible optical networks, higher and higher requirements are put on the identification of signal modulation formats.
Usually, a neural network is used to identify the signal modulation format, that is, a characteristic parameter obtained by sampling and extracting a signal is input into the trained neural network to obtain the signal modulation format. However, the existing method for identifying the signal modulation format needs to use electro-optical or photoelectric conversion during sampling and parameter extraction, and the conversion process takes much time. Moreover, the speed of the conventional processor is relatively slow when the processor processes the matrix operation in the neural network. In addition, the power consumption of this method is also high.
Therefore, the existing method for identifying the signal modulation format has the disadvantages of low speed and high power consumption.
Disclosure of Invention
The embodiment of the invention aims to provide a signal modulation format identification method and device based on a photonic neural network, electronic equipment and a storage medium, so as to solve the problems of low speed and high power consumption of the signal modulation format identification method in the prior art.
The embodiment of the invention provides a signal modulation format identification method based on a photon neural network, which comprises the following steps:
acquiring a characteristic signal of a signal to be identified, wherein the characteristic signal is obtained after the signal to be identified is sampled;
inputting the characteristic signal into a photonic chip, wherein the photonic chip is used for completing linear operation in a neural network algorithm for signal modulation format recognition, and the neural network algorithm is completed in advance according to signal training of a plurality of known modulation formats;
and acquiring an output result of the photonic chip after linear operation, and performing nonlinear operation in the neural network algorithm based on the output result to obtain a modulation format of the signal to be identified.
Optionally, the photonic chip includes a mach-zehnder interferometer;
determining the parameter value of the Mach-Zehnder interferometer by adopting the following method:
determining a transmission matrix in a neural network algorithm for signal modulation format identification;
decomposing the transmission matrix into a unitary matrix and a diagonal matrix, wherein the product of the unitary matrix and the diagonal matrix is the transmission matrix;
constructing a plurality of matrixes to be solved, wherein the structures of the matrixes to be solved are the same as the unitary matrix, elements at four specific positions in the matrixes to be solved are to be solved, except that the elements at other positions of the four specific positions are the same as the elements at other positions in the unit matrix, and the elements at the four specific positions contain phase parameters to be solved;
based on the unitary matrix right-multiplying the multiple matrixes to be solved to obtain a relational expression of a unit matrix, solving the multiple matrixes to be solved, and taking the result as an auxiliary matrix;
and determining the parameter value of the Mach-Zehnder interferometer corresponding to each auxiliary matrix based on the one-to-one correspondence relationship between the auxiliary matrix and the parameter value of the Mach-Zehnder interferometer.
Optionally, the constructing a plurality of matrices to be solved includes:
for an N-dimensional unitary matrix U (N), an N-dimensional matrix T to be solved is constructedm,nWherein m and n are positive integers, and n is more than or equal to 1<m≤N；
Figure BDA0001628448030000021
Wherein the matrix T to be solvedm,nThe element of the nth row and the nth column of (1) is exp (i phi)m,n)sin(θm,n/2), the element in the n-th row and m-th column is cos (theta)m,n/2), line mThe element in the nth column is exp (i φ)m,n)cos(θm,n/2), the element in the mth row and mth column is sin (θ)m,n/2)，φm,n、θm,nFor the matrix T to be solvedm,nThe phase parameter of (2);
the right-multiplying the unitary matrix by the multiple matrixes to be solved to obtain a relational expression of an identity matrix, solving the multiple matrixes to be solved, and taking the result as an auxiliary matrix, wherein the relational expression comprises the following steps:
step 1: for the N-dimensional unitary matrix U (N), a corresponding matrix T to be solved is calculated according to the following formulaN,N-1、TN,N-2…TN,1And remember TN,N-1、TN,N-2…TN,1Is TN；
Figure BDA0001628448030000031
Step 2: calculating the matrix T to be solvedN,N-1、TN,N-2…TN,1Then, calculating U (N-1) according to the following formula, and returning to the step 1 after calculating to obtain U (N-1) until calculating U (1);
Figure BDA0001628448030000032
wherein U (N-1) is an N-1-dimensional unitary matrix corresponding to the N-dimensional unitary matrix U (N);
step 3, obtaining T in the calculation processN、TN-1、TN-2…T2Determining an auxiliary matrix which is the N-dimensional unitary matrix U (N).
Optionally, determining the parameter value of the mach-zehnder interferometer corresponding to each auxiliary matrix based on the one-to-one correspondence relationship between the auxiliary matrix and the parameter value of the mach-zehnder interferometer includes:
and determining the phase parameter value in the auxiliary matrix as the parameter value of the Mach-Zehnder interferometer corresponding to the auxiliary matrix.
The embodiment of the invention provides a signal modulation format recognition device based on a photon neural network, which comprises:
the device comprises a characteristic signal acquisition module, a characteristic signal acquisition module and a signal recognition module, wherein the characteristic signal acquisition module is used for acquiring a characteristic signal of a signal to be recognized, and the characteristic signal is obtained after the signal to be recognized is sampled;
the input module is used for inputting the characteristic signals into a photonic chip, the photonic chip is used for completing linear operation in a neural network algorithm for signal modulation format recognition, and the neural network algorithm is completed according to signal training of a plurality of known modulation formats in advance;
and the output result acquisition module is used for acquiring an output result of the photonic chip after linear operation, and performing nonlinear operation in the neural network algorithm based on the output result to obtain a modulation format of the signal to be identified.
Optionally, the photonic chip includes a mach-zehnder interferometer, and the apparatus further includes:
an interferometer parameter value determining module, configured to determine a parameter value of the mach-zehnder interferometer;
the interferometer parameter value determination module comprises:
the transmission matrix determining module is used for determining a transmission matrix in a neural network algorithm for signal modulation format recognition;
a matrix decomposition module, configured to decompose the transmission matrix into a unitary matrix and a diagonal matrix, where a product of the unitary matrix and the diagonal matrix is the transmission matrix;
the matrix construction module is used for constructing a plurality of matrixes to be solved, the structures of the matrixes to be solved are the same as the unitary matrix, elements at four specific positions in the matrixes to be solved are to be solved, the elements at other positions except the four specific positions are the same as the elements at other positions in the unit matrix, and the elements at the four specific positions contain phase parameters to be solved;
the matrix solving module is used for obtaining a relational expression of a unit matrix by right multiplying the unitary matrix by the matrixes to be solved, solving the matrixes to be solved, and taking the result as an auxiliary matrix;
and the parameter value determining module is used for determining the parameter value of the Mach-Zehnder interferometer corresponding to each auxiliary matrix based on the one-to-one correspondence relationship between the auxiliary matrix and the parameter value of the Mach-Zehnder interferometer.
Optionally, the matrix constructing module is specifically configured to:
for an N-dimensional unitary matrix U (N), an N-dimensional matrix T to be solved is constructedm,nWherein m and n are positive integers, and n is more than or equal to 1<m≤N；
Figure BDA0001628448030000051
Wherein the matrix T to be solvedm,nThe element of the nth row and the nth column of (1) is exp (i phi)m,n)sin(θm,n/2), the element in the n-th row and m-th column is cos (theta)m,n/2), the element in the mth row and nth column is exp (i φ)m,n)cos(θm,n/2), the element in the mth row and mth column is sin (θ)m,n/2)，φm,n、θm,nFor the matrix T to be solvedm,nThe phase parameter of (2);
the matrix solving module is, in particular for,
step 1: for the N-dimensional unitary matrix U (N), a corresponding matrix T to be solved is calculated according to the following formulaN,N-1、TN,N-2…TN,1And remember TN,N-1、TN,N-2…TN,1Is TN；
Figure BDA0001628448030000052
Step 2: calculating the matrix T to be solvedN,N-1、TN,N-2…TN,1Then, calculating U (N-1) according to the following formula, and returning to the step 1 after calculating to obtain U (N-1) until calculating U (1);
Figure BDA0001628448030000061
wherein U (N-1) is an N-1-dimensional unitary matrix corresponding to the N-dimensional unitary matrix U (N);
step 3, obtaining T in the calculation processN、TN-1、TN-2…T2Determining an auxiliary matrix which is the N-dimensional unitary matrix U (N).
Optionally, the parameter value determining module is specifically configured to determine a phase parameter value in the auxiliary matrix as a parameter value of a mach-zehnder interferometer corresponding to the auxiliary matrix.
Therefore, the signal modulation format recognition method and device based on the photonic neural network provided by the embodiment of the invention use the photonic chip to complete the linear operation in the neural network algorithm for signal modulation format recognition trained in advance, and then perform the nonlinear operation in the neural network algorithm on the basis of the linear operation result, so as to obtain the modulation format of the signal to be recognized. Therefore, a processor is not needed to process matrix operation contained in a linear operation part in the neural network, the speed of signal modulation format identification is improved, and the energy consumption is low when a photonic chip is used for operation, so that the power consumption of the whole system is reduced by using the method.
The embodiment of the invention also provides electronic equipment which comprises a processor, a communication interface, a memory, a communication bus and a photonic chip, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus; the processor is connected with the photonic chip through a photoelectric conversion interface;
the memory is used for storing a computer program;
the processor is used for realizing any one of the method steps when executing the program stored in the memory;
and the photonic chip is used for finishing linear operation in the neural network algorithm for signal modulation format identification and sending an operation result to the processor.
An embodiment of the present invention further provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the computer program implements any of the above method steps.
Of course, it is not necessary for any product or method of practicing the invention to achieve all of the above-described advantages at the same time.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a signal modulation format recognition method based on a photonic neural network according to an embodiment of the present invention;
FIG. 2 is a flow chart for determining the values of Mach-Zehnder interferometer parameters provided by an embodiment of the present invention;
FIG. 3 is a schematic diagram of an optical implementation of a unitary matrix using a Mach-Zehnder interferometer according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a signal modulation format recognition apparatus based on a photonic neural network according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to solve the problems of low speed and high power consumption of a signal modulation format identification method in the prior art, the embodiment of the invention provides a signal modulation format identification method and device based on a photonic neural network, electronic equipment and a storage medium.
First, a description is given to a signal modulation format recognition method based on a photonic neural network according to an embodiment of the present invention, as shown in fig. 1, the method may include the following steps:
step S101: acquiring a characteristic signal of a signal to be identified, wherein the characteristic signal is obtained after sampling the signal to be identified;
in the embodiment of the invention, in the process of identifying the modulation format of the signal, the signal to be identified can be sampled first to obtain the characteristic signal. The sampling process may be performed by methods in the prior art, and is not limited thereto.
Step S102: inputting the characteristic signal into a photonic chip, wherein the photonic chip is used for completing linear operation in a neural network algorithm for identifying a signal modulation format, and the neural network algorithm is completed according to training of a plurality of signals with known modulation formats in advance;
in the embodiment of the present invention, a neural network algorithm for signal modulation format recognition may be trained in advance according to a method in the prior art, which is not limited to this.
After the trained neural network algorithm for signal modulation format recognition is obtained, transmission matrixes of a linear operation part in the neural network algorithm can be determined, parameters of optical devices are determined by the transmission matrixes, the optical devices are integrated into a photonic chip, and then the photonic chip can be used for completing the linear operation in the neural network algorithm.
A method for determining parameters of an optical device using a transmission matrix can be seen below.
Step S103: and acquiring an output result of the photonic chip after linear operation, and performing nonlinear operation in a neural network algorithm based on the output result to obtain a modulation format of the signal to be identified.
In the embodiment of the invention, the integrated photonic chip can be connected with a processor of the electronic device through the photoelectric conversion interface, the output result of the photonic chip after linear operation can be received by the processor, and the processor continues to perform nonlinear operation in the neural network algorithm based on the output result, so that the modulation format of the signal to be identified can be obtained. Of course, the non-linear operation may be performed by a processor or a specific computing unit, which is not limited in the embodiment of the present invention.
Therefore, according to the signal modulation format recognition method and device based on the photonic neural network provided by the embodiment of the invention, the linear operation in the pre-trained neural network algorithm for signal modulation format recognition is completed by using the photonic chip, and the processor performs the nonlinear operation in the neural network algorithm on the basis of the linear operation result, so that the modulation format of the signal to be recognized can be obtained. Therefore, a processor is not needed to process matrix operation contained in a linear operation part in the neural network, the speed of signal modulation format identification is improved, and the energy consumption is low when a photonic chip is used for operation, so that the power consumption of the whole system is reduced by using the method.
In the embodiment of the present invention, the transmission matrix in the neural network may be implemented by using optical devices, namely, a mach-zehnder interferometer, a gain device, an attenuator, and a phase shifter, wherein the matrix representation form of the mach-zehnder interferometer is as follows:
Figure BDA0001628448030000091
where theta and phi are parameters of the mach-zehnder interferometer.
Referring to fig. 2, fig. 2 is a flow chart for determining a parameter value of a mach-zehnder interferometer according to an embodiment of the present invention, which may include the following steps:
step S201: determining a transmission matrix in a neural network algorithm for signal modulation format identification;
in the embodiment of the invention, a trained neural network algorithm for signal modulation format recognition can be obtained in advance, and then transmission matrixes included in the algorithm can be determined, and element values of all positions in the matrixes can be determined.
Step S202: the transmission matrix is decomposed into a unitary matrix and a diagonal matrix, and a product of the unitary matrix and the diagonal matrix is the transmission matrix.
For a transmission matrix M, the transmission matrix can be decomposed into a unitary matrix and a diagonal matrix using the following formula:
M＝UΣV*
wherein, U, V*Being a unitary matrix, the unitary matrix can be represented using a mach-zehnder interferometer. Σ is a diagonal matrix, which can be represented in a photonic chip using a gain or attenuator. The method for decomposing the transmission matrix into the unitary matrix and the diagonal matrix may adopt a singular value decomposition method in the prior art, and may also adopt other methods, which is not limited in the embodiment of the present invention.
Step S203: constructing a plurality of matrixes to be solved, wherein the structures of the matrixes to be solved are the same as the unitary matrix, elements at four specific positions in the matrixes to be solved are to be solved, the elements at other positions except the four specific positions are the same as the elements at other positions in the unit matrix, and the elements at the four specific positions contain phase parameters to be solved;
step S204: obtaining a relational expression of an identity matrix by right multiplying a plurality of matrixes to be solved based on a unitary matrix, solving the plurality of matrixes to be solved, and taking the result as an auxiliary matrix;
in the embodiment of the invention, in the photonic chip, a plurality of mach-zehnder interferometers can be used to represent a unitary matrix, and in order to determine parameters of the plurality of mach-zehnder interferometers corresponding to each unitary matrix, each matrix to be solved can be solved by constructing the matrix to be solved and carrying out product operation on the unitary matrix and right multiplying the unitary matrix by the plurality of matrices to be solved to obtain a relational expression of the unit matrix.
For an N-dimensional unitary matrix U (N), a structure can be constructed
Figure BDA0001628448030000101
The product operation is carried out on the matrix to be solved and the unitary matrix, and the matrix to be solved can be recorded as Tm,nWherein m and n are positive integers, and n is more than or equal to 1<m is less than or equal to N; then there is a total of
Figure BDA0001628448030000102
And (5) solving the matrix.
For example, for a 4-dimensional unitary matrix U (4), T can be constructed4,3、T4,2、T4,1、T3,2、T3,1、T2,1And 6 matrixes to be solved are totally 6, namely, for the 4-dimensional unitary matrix, parameters in the 6 matrixes to be solved can be solved firstly, and the 4-dimensional unitary matrix is represented in a photonic chip by using 6 Mach-Zehnder interferometers after the solution.
Matrix T to be solvedm,nCan be expressed as
Figure BDA0001628448030000103
Wherein, the matrix T to be solvedm,nThe element of the nth row and the nth column of (1) is exp (i phi)m,n)sin(θm,n/2), the element in the n-th row and m-th column is cos (theta)m,n/2), the element in the mth row and nth column is exp (i φ)m,n)cos(θm,n/2), the element in the mth row and mth column is sin (θ)m,n/2)，φm,n、θm,nFor the matrix T to be solvedm,nThe phase parameter of (2);
for example, for the matrix T to be solved corresponding to the above-mentioned 4-dimensional unitary matrix3,2Then, only four elements in the 2 nd row, 2 nd column, 2 nd row, 3 rd column, 3 rd row, 2 nd column, and 3 rd row, 3 rd column in the matrix to be solved are unknown, and the element values at other positions are the same as the element values at the other positions in the 4-dimensional unit matrix. And the elements at the four positions are exp (i phi)3,2)sin(θ3,2/2)，cos(θ3,2/2)，exp(iφ3,2)cos(θ3,2/2)，sin(θ3,2/2) wherein phi3,2And theta3,2Namely the matrix T to be solved3,2The phase parameter of (2);
is constructed according to the above rules
Figure BDA0001628448030000111
After the matrixes to be solved are obtained, the unitary matrixes can be used for sequentially multiplying the matrixes to be solved right, and the phase in each matrix to be solved is calculated according to the product relationA bit parameter.
For N-dimensional unitary matrix U (N), right multiplying T in sequenceN,N-1、TN,N-2…TN,1Then, a matrix in which the last element in the last row is 0 except the last element is not zero can be obtained, that is, the following relation is satisfied:
Figure BDA0001628448030000112
will TN,N-1、TN,N-2…TN,1Substituting the operation formula into the relational expression to extract the operation formula about the last row matrix, and obtaining the following formula:
Figure BDA0001628448030000113
as can be seen from the above relation, on the basis that U (N) is known, theta can be calculated from the relationN,N-1，θN,N-2…θN,1And phiN,N-1，φN,N-2…φN,1Which is the matrix T to be solvedN,N-1、TN,N-2…TN,1After determining the phase parameters, the matrix T to be solvedN,N-1、TN,N-2…TN,1The element values for each position in the array can be determined.
In the calculation of the matrix T to be solvedN,N-1、TN,N-2…TN,1Then, U (N-1) is calculated as follows:
Figure BDA0001628448030000114
where U (N-1) is a unitary matrix of dimension N-1, then T can be calculated from the unitary matrix of dimension N as described aboveN,N-1、TN,N-2…TN,1The same algorithm, calculating T from the N-1 dimensional unitary matrix U (N-1)N-1,N-2、TN-1,N-3…TN-1,1And because a new unitary matrix with one dimensionality reduced by one can be obtained after each calculation, other matrixes to be solved can be calculated according to the same method.
T obtained in the calculation processN、TN-1、TN-2…T2Determining auxiliary matrix of N-dimensional unitary matrix U (N), total
Figure BDA0001628448030000115
And auxiliary matrixes, each auxiliary matrix corresponding to the parameter values of the Mach-Zehnder interferometer one by one.
Step S205: and determining the parameter value of the Mach-Zehnder interferometer corresponding to each auxiliary matrix based on the one-to-one correspondence relationship between the auxiliary matrix and the parameter value of the Mach-Zehnder interferometer.
Each auxiliary matrix comprises phase parameters phi and theta, and after the auxiliary matrix is solved, the values of the phase parameters contained in the auxiliary matrix can be determined. Since the auxiliary matrices have a one-to-one correspondence with the parameter values of the mach-zehnder interferometers, the phase parameter value of each auxiliary matrix can be determined as the parameter value of the mach-zehnder interferometer to which the auxiliary matrix corresponds.
In the photonic chip, the mach-zehnder interferometers corresponding to the unitary matrix may be connected according to a certain rule to represent the unitary matrix.
Referring to fig. 3, fig. 3 is a schematic diagram of an embodiment of the present invention for optically implementing a unitary matrix by using a mach-zehnder interferometer, fig. 3 is an exemplary 5-dimensional unitary matrix, the 5-dimensional unitary matrix can be optically implemented by using 10 mach-zehnder interferometers, and parameter values of the 10 mach-zehnder interferometers can be predetermined. Phase shifters may be used to ensure that the phase of each mach-zehnder interferometer port is uniform.
Therefore, in the embodiment of the invention, the parameters of the optical device can be determined according to the transmission matrix in the neural network, then the optical device is connected according to a certain rule, the transmission matrix can be realized, and finally the linear operation in the neural network for signal modulation format identification can be completed by using the photonic chip integrated with the optical device, so that a processor is not needed to complete the linear operation, therefore, the speed of the signal modulation format identification method is improved, and the power consumption is reduced.
The embodiment of the invention also provides a signal modulation format recognition device based on the photonic neural network, which can be seen in fig. 4 and comprises the following modules:
the characteristic signal acquiring module 401 is configured to acquire a characteristic signal of a signal to be identified, where the characteristic signal is obtained after sampling the signal to be identified;
an input module 402, configured to input the characteristic signal into a photonic chip, where the photonic chip is configured to perform linear operation in a neural network algorithm for signal modulation format recognition, and the neural network algorithm is performed in advance according to signal training of a plurality of known modulation formats;
the output result obtaining module 403 is configured to obtain an output result obtained after the photonic chip performs linear operation, and perform nonlinear operation in a neural network algorithm based on the output result to obtain a modulation format of the signal to be identified.
In the embodiment of the present invention, on the basis of the apparatus shown in fig. 4, the apparatus may further include:
the interferometer parameter determining module is used for determining the parameter value of the Mach-Zehnder interferometer;
an interferometer parameter determination module may include:
the transmission matrix determining module is used for determining a transmission matrix in a neural network algorithm for signal modulation format recognition;
the matrix decomposition module is used for decomposing the transmission matrix into a unitary matrix and a diagonal matrix, and the product of the unitary matrix and the diagonal matrix is the transmission matrix;
the matrix construction module is used for constructing a plurality of matrixes to be solved, the structures of the matrixes to be solved are the same as the unitary matrix, elements at four specific positions in the matrixes to be solved are to be solved, the elements at other positions except the four specific positions are the same as the elements at the other positions in the unit matrix, and the elements at the four specific positions contain phase parameters to be solved;
the matrix solving module is used for obtaining a relational expression of the unit matrix by right multiplying the unitary matrix by the matrixes to be solved, solving the matrixes to be solved, and taking the result as an auxiliary matrix;
and the parameter value determining module is used for determining the parameter value of the Mach-Zehnder interferometer corresponding to each auxiliary matrix based on the one-to-one correspondence relationship between the auxiliary matrix and the parameter value of the Mach-Zehnder interferometer.
In the embodiment of the present invention, the matrix construction module may be specifically configured to:
for an N-dimensional unitary matrix U (N), an N-dimensional matrix T to be solved is constructedm,nWherein m and n are positive integers, and n is more than or equal to 1<m≤N；
Figure BDA0001628448030000131
Wherein, the matrix T to be solvedm,nThe element of the nth row and the nth column of (1) is exp (i phi)m,n)sin(θm,n/2), the element in the n-th row and m-th column is cos (theta)m,n/2), the element in the mth row and nth column is exp (i φ)m,n)cos(θm,n/2), the element in the mth row and mth column is sin (θ)m,n/2)，φm,n、θm,nFor the matrix T to be solvedm,nThe phase parameter of (2);
the matrix solving module may be specifically configured to:
step 1: for the N-dimensional unitary matrix U (N), the corresponding matrix T to be solved is calculated according to the following formulaN,N-1、TN,N-2…TN,1And remember TN,N-1、TN,N-2…TN,1Is TN；
Figure BDA0001628448030000141
Step 2: in the calculation of the matrix T to be solvedN,N-1、TN,N-2…TN,1Then, calculating U (N-1) according to the following formula, and returning to the step 1 after calculating to obtain U (N-1) until calculating U (1);
Figure BDA0001628448030000142
wherein, U (N-1) is an N-1 dimensional unitary matrix corresponding to an N-dimensional unitary matrix U (N);
step 3, obtaining T in the calculation processN、TN-1、TN-2…T2An auxiliary matrix is determined which is an N-dimensional unitary matrix u (N).
And the parameter value determining module is specifically used for determining the phase parameter value in the auxiliary matrix as the parameter value of the Mach-Zehnder interferometer corresponding to the auxiliary matrix.
In the embodiment of the invention, the parameters of the optical device can be determined according to the transmission matrix in the neural network, then the optical device is connected according to a certain rule, the transmission matrix can be realized, and finally the linear operation in the neural network for signal modulation format identification can be completed by using the photonic chip integrated with the optical device, so that a processor is not needed to complete the linear operation, thereby improving the speed of the signal modulation format identification method and reducing the power consumption.
Correspondingly, an embodiment of the present invention provides an electronic device, which may refer to fig. 5, and includes a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete mutual communication through the communication bus;
a memory for storing a computer program;
and the processor is used for realizing the steps of the method when executing the program stored in the memory.
The communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.
The communication interface is used for communication between the electronic equipment and other equipment.
The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the processor.
The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component.
The embodiment of the invention also provides a computer readable storage medium, wherein a computer program is stored in the storage medium, and the computer program is executed by a processor to realize any one of the method steps.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the apparatus embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.

Claims (7)

1. A signal modulation format identification method based on a photonic neural network is characterized by comprising the following steps:
acquiring a characteristic signal of a signal to be identified, wherein the characteristic signal is obtained after the signal to be identified is sampled;
inputting the characteristic signal into a photonic chip, wherein the photonic chip is used for completing linear operation in a neural network algorithm for signal modulation format recognition, and the neural network algorithm is completed in advance according to signal training of a plurality of known modulation formats; the photonic chip is formed by connecting optical devices, and the optical devices comprise Mach-Zehnder interferometers, gain devices, attenuators and phase shifters; the parameters of the optical device are determined by a transmission matrix of a linear operation part in the neural network algorithm;
acquiring an output result of the photonic chip after linear operation, and performing nonlinear operation in the neural network algorithm based on the output result to obtain a modulation format of the signal to be identified;
determining the parameter value of the Mach-Zehnder interferometer by adopting the following method:
determining a transmission matrix in a neural network algorithm for signal modulation format identification;
decomposing the transmission matrix into a unitary matrix and a diagonal matrix, wherein the product of the unitary matrix and the diagonal matrix is the transmission matrix;
constructing a plurality of matrixes to be solved, wherein the structures of the matrixes to be solved are the same as the unitary matrix, elements at four specific positions in the matrixes to be solved are to be solved, except that the elements at other positions of the four specific positions are the same as the elements at other positions in the unit matrix, and the elements at the four specific positions contain phase parameters to be solved;
based on the unitary matrix right-multiplying the multiple matrixes to be solved to obtain a relational expression of a unit matrix, solving the multiple matrixes to be solved, and taking the result as an auxiliary matrix;
and determining the parameter value of the Mach-Zehnder interferometer corresponding to each auxiliary matrix based on the one-to-one correspondence relationship between the auxiliary matrix and the parameter value of the Mach-Zehnder interferometer.
2. The method of claim 1, wherein constructing a plurality of matrices to be solved comprises:
for an N-dimensional unitary matrix U (N), an N-dimensional matrix T to be solved is constructedm,nWherein m and N are positive integers, and N is more than or equal to 1 and less than or equal to m and less than or equal to N;
Figure FDA0002239367990000021
wherein the matrix T to be solvedm,nThe element of the nth row and the nth column of (1) is exp (i phi)m,n)sin(θm,n/2), the element in the n-th row and m-th column is cos (theta)m,n/2), the element in the mth row and nth column is exp (i φ)m,n)cos(θm,n/2), the element in the mth row and mth column is sin (θ)m,n/2)，φm,n、θm,nFor the matrix T to be solvedm,nThe phase parameter of (2);
the right-multiplying the unitary matrix by the multiple matrixes to be solved to obtain a relational expression of an identity matrix, solving the multiple matrixes to be solved, and taking the result as an auxiliary matrix, wherein the relational expression comprises the following steps:
step 1: for the N-dimensional unitary matrix U (N), a corresponding matrix T to be solved is calculated according to the following formulaN,N-1、TN,N-2…TN,1And remember TN,N-1、TN,N-2…TN,1Is TN；
Figure FDA0002239367990000022
Step 2: at the position of calculatingThe matrix T to be solvedN,N-1、TN,N-2…TN,1Then, calculating U (N-1) according to the following formula, and returning to the step 1 after calculating to obtain U (N-1) until calculating U (1);
Figure FDA0002239367990000031
wherein, U (N-1) is an N-1 dimensional unitary matrix corresponding to the N dimensional unitary matrix U (N), α is an adjusting phase;
step 3, obtaining T in the calculation processN、TN-1、TN-2…T2Determining an auxiliary matrix which is the N-dimensional unitary matrix U (N).
3. The method of claim 2, wherein determining the parameter values of the mach-zehnder interferometer corresponding to each of the auxiliary matrices based on the one-to-one correspondence of the auxiliary matrices to the parameter values of the mach-zehnder interferometer comprises:
and determining the phase parameter value in the auxiliary matrix as the parameter value of the Mach-Zehnder interferometer corresponding to the auxiliary matrix.
4. An apparatus for identifying a signal modulation format based on a photonic neural network, the apparatus comprising:
the device comprises a characteristic signal acquisition module, a characteristic signal acquisition module and a signal recognition module, wherein the characteristic signal acquisition module is used for acquiring a characteristic signal of a signal to be recognized, and the characteristic signal is obtained after the signal to be recognized is sampled;
the input module is used for inputting the characteristic signals into a photonic chip, the photonic chip is used for completing linear operation in a neural network algorithm for signal modulation format recognition, and the neural network algorithm is completed according to signal training of a plurality of known modulation formats in advance; the photonic chip is formed by connecting optical devices, and the optical devices comprise Mach-Zehnder interferometers, gain devices, attenuators and phase shifters; the parameters of the optical device are determined by a transmission matrix of a linear operation part in the neural network algorithm;
the output result acquisition module is used for acquiring an output result of the photonic chip after linear operation and carrying out nonlinear operation in the neural network algorithm based on the output result to obtain a modulation format of the signal to be identified;
the device further comprises:
an interferometer parameter value determining module, configured to determine a parameter value of the mach-zehnder interferometer;
the interferometer parameter value determination module comprises:
the transmission matrix determining module is used for determining a transmission matrix in a neural network algorithm for signal modulation format recognition;
a matrix decomposition module, configured to decompose the transmission matrix into a unitary matrix and a diagonal matrix, where a product of the unitary matrix and the diagonal matrix is the transmission matrix;
the matrix construction module is used for constructing a plurality of matrixes to be solved, the structures of the matrixes to be solved are the same as the unitary matrix, elements at four specific positions in the matrixes to be solved are to be solved, the elements at other positions except the four specific positions are the same as the elements at other positions in the unit matrix, and the elements at the four specific positions contain phase parameters to be solved;
the matrix solving module is used for obtaining a relational expression of a unit matrix by right multiplying the unitary matrix by the matrixes to be solved, solving the matrixes to be solved, and taking the result as an auxiliary matrix;
and the parameter value determining module is used for determining the parameter value of the Mach-Zehnder interferometer corresponding to each auxiliary matrix based on the one-to-one correspondence relationship between the auxiliary matrix and the parameter value of the Mach-Zehnder interferometer.
5. The apparatus according to claim 4, wherein the matrix construction module is specifically configured to:
for an N-dimensional unitary matrix U (N), an N-dimensional matrix T to be solved is constructedm,nWherein m and N are positive integers, and N is more than or equal to 1 and less than or equal to m and less than or equal to N;
Figure FDA0002239367990000041
wherein the matrix T to be solvedm,nThe element of the nth row and the nth column of (1) is exp (i phi)m,n)sin(θm,n/2), the element in the n-th row and m-th column is cos (theta)m,n/2), the element in the mth row and nth column is exp (i φ)m,n)cos(θm,n/2), the element in the mth row and mth column is sin (θ)m,n/2)，φm,n、θm,nFor the matrix T to be solvedm,nThe phase parameter of (2);
the matrix solving module is, in particular for,
step 1: for the N-dimensional unitary matrix U (N), a corresponding matrix T to be solved is calculated according to the following formulaN,N-1、TN,N-2…TN,1And remember TN,N-1、TN,N-2…TN,1Is TN；
Figure FDA0002239367990000051
Step 2: calculating the matrix T to be solvedN,N-1、TN,N-2…TN,1Then, calculating U (N-1) according to the following formula, and returning to the step 1 after calculating to obtain U (N-1) until calculating U (1);
Figure FDA0002239367990000052
wherein, U (N-1) is an N-1 dimensional unitary matrix corresponding to the N dimensional unitary matrix U (N), α is an adjusting phase;
step 3, obtaining T in the calculation processN、TN-1、TN-2…T2Determining an auxiliary matrix which is the N-dimensional unitary matrix U (N).
6. The apparatus according to claim 5, wherein the parameter value determining module is configured to determine the phase parameter value in the auxiliary matrix as a parameter value of a Mach-Zehnder interferometer corresponding to the auxiliary matrix.
7. The electronic equipment is characterized by comprising a processor, a communication interface, a memory, a communication bus and a photonic chip, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus; the processor is connected with the photonic chip through a photoelectric conversion interface;
the memory is used for storing a computer program;
the processor, when executing the program stored in the memory, implementing the method steps of claims 1-3;
and the photonic chip is used for finishing linear operation in the neural network algorithm for signal modulation format identification and sending an operation result to the processor.
CN201810332866.0A 2018-04-13 2018-04-13 Signal modulation format identification method and device based on photon neural network Active CN108599865B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810332866.0A CN108599865B (en) 2018-04-13 2018-04-13 Signal modulation format identification method and device based on photon neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810332866.0A CN108599865B (en) 2018-04-13 2018-04-13 Signal modulation format identification method and device based on photon neural network

Publications (2)

Publication Number Publication Date
CN108599865A CN108599865A (en) 2018-09-28
CN108599865B true CN108599865B (en) 2020-04-28

Family

ID=63622571

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810332866.0A Active CN108599865B (en) 2018-04-13 2018-04-13 Signal modulation format identification method and device based on photon neural network

Country Status (1)

Country Link
CN (1) CN108599865B (en)

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109376855B (en) * 2018-12-14 2021-04-06 中国科学院计算技术研究所 Optical neuron structure and neural network processing system comprising same
CN109784486B (en) * 2018-12-26 2021-04-23 中国科学院计算技术研究所 Optical neural network processor and training method thereof
CN109871871B (en) * 2019-01-16 2021-08-27 南方科技大学 Image identification method and device based on optical neural network structure and electronic equipment
CN112396171A (en) * 2019-08-15 2021-02-23 杭州智芯科微电子科技有限公司 Artificial intelligence computing chip and signal processing system
CN110597756B (en) * 2019-08-26 2023-07-25 光子算数(北京)科技有限责任公司 Calculation circuit and data operation method
CN110503196A (en) * 2019-08-26 2019-11-26 光子算数(北京)科技有限责任公司 A kind of photon neural network chip and data processing system
CN110505021B (en) * 2019-08-26 2021-04-09 光子算数(北京)科技有限责任公司 Optical communication device, optical communication system and optical signal processing method
CN110852431B (en) * 2019-11-05 2023-03-31 联合微电子中心有限责任公司 Digital signal modulation method of photon artificial intelligence computing chip
CN111093123B (en) * 2019-12-09 2020-12-18 华中科技大学 Flexible optical network time domain equalization method and system based on composite neural network
CN110874206A (en) * 2020-01-19 2020-03-10 光子算数(北京)科技有限责任公司 Data processing method and device based on optical chip, storage medium and electronic equipment
CN112232503B (en) * 2020-06-09 2021-11-19 联合微电子中心有限责任公司 Computing device, computing method, and computing system
CN111753977B (en) * 2020-06-30 2024-01-02 中国科学院半导体研究所 Optical neural network convolution layer chip, convolution calculation method and electronic equipment
CN112232504B (en) * 2020-09-11 2024-06-21 联合微电子中心有限责任公司 Photonic neural network
CN113110493B (en) * 2021-05-07 2022-09-30 北京邮电大学 Path planning equipment and path planning method based on photonic neural network
CN113592069A (en) * 2021-07-20 2021-11-02 东南大学 Photon neural network for four-input logic operation

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0626937A (en) * 1992-07-08 1994-02-04 Advantest Corp Measuring method of optical pulse waveform
CN101097282A (en) * 2006-06-30 2008-01-02 中国石油天然气集团公司 Optical fiber safety early warning signal identification system
CN102737462A (en) * 2012-06-29 2012-10-17 天津大学 Optical fiber distributed disturbance and video linkage long-distance circumference security monitoring system

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3213081A4 (en) * 2014-10-30 2018-08-15 Sightline Innovation Inc. System, method and apparatus for pathogen detection

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0626937A (en) * 1992-07-08 1994-02-04 Advantest Corp Measuring method of optical pulse waveform
CN101097282A (en) * 2006-06-30 2008-01-02 中国石油天然气集团公司 Optical fiber safety early warning signal identification system
CN102737462A (en) * 2012-06-29 2012-10-17 天津大学 Optical fiber distributed disturbance and video linkage long-distance circumference security monitoring system

Also Published As

Publication number Publication date
CN108599865A (en) 2018-09-28

Similar Documents

Publication Publication Date Title
CN108599865B (en) Signal modulation format identification method and device based on photon neural network
US11042782B2 (en) Topic-guided model for image captioning system
CN110798417B (en) Signal modulation identification method and device based on cyclic residual error network
CN109871871B (en) Image identification method and device based on optical neural network structure and electronic equipment
Zhang et al. Modulation classification method for frequency modulation signals based on the time–frequency distribution and CNN
US11748595B2 (en) Convolution acceleration operation method and apparatus, storage medium and terminal device
CN110932809B (en) Fiber channel model simulation method, device, electronic equipment and storage medium
US11294985B2 (en) Efficient analog in-memory matrix multiplication processor
US20220172052A1 (en) Machine learning model training using an analog processor
CN109753987B (en) File recognition method and feature extraction method
CN114520736B (en) Internet of things security detection method, device, equipment and storage medium
CN112767927A (en) Method, device, terminal and storage medium for extracting voice features
CN112748483A (en) Air temperature forecast deviation correction method and device based on deep learning
Wei et al. Finite‐time stochastic stabilisation of Markovian jump non‐linear quadratic systems with partially known transition probabilities
CN111458676B (en) Direction-of-arrival estimation method and device based on cascaded neural network
Zhao et al. Inference for modulated stationary processes
CN114444657A (en) Image processing method, system, equipment and readable storage medium
CN113705774A (en) Optical circuit construction method, optical circuit, optical signal processing method and device
Matni et al. Low-rank and low-order decompositions for local system identification
CN111931938B (en) Cyclic neural network reasoning operation acceleration system and method based on structured sparsity
CN107644004A (en) A kind of Digital Signal Processing Method And Device based on discrete fractional Brownian random field quick calculation method
Boudjenouia et al. Robust, blind multichannel image identification and restoration using stack decoder
CN111078821B (en) Dictionary setting method, dictionary setting device, medium and electronic equipment
CN114419341A (en) Convolutional neural network image identification method based on transfer learning improvement
CN116318470B (en) Method and device for estimating communication interference signal power under non-Gaussian noise

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