CN117335879A

CN117335879A - Optical signal processing device and related method

Info

Publication number: CN117335879A
Application number: CN202210726895.1A
Authority: CN
Inventors: 程铭; 方涛; 吴彤宇; 董晓文
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2022-06-24
Filing date: 2022-06-24
Publication date: 2024-01-02

Abstract

The application discloses an optical signal processing device and a related method, wherein the device comprises a Brillouin optical signal reading and writing unit: the Brillouin optical signal reading and writing unit is used for receiving a writing optical signal, a first reading optical signal and a second reading optical signal, wherein the writing optical signal and the second reading optical signal bear a first signal to be processed; the Brillouin optical signal reading and writing unit is further used for realizing a linear dimension-increasing function according to the writing optical signal and the first reading optical signal to obtain a first output optical signal; the brillouin optical signal reading and writing unit is further configured to implement a nonlinear mapping function according to the writing optical signal and the second reading optical signal, and obtain a second output optical signal. According to the embodiment of the application, the Brillouin optical signal read-write unit is arranged, so that multiple functions in the NGRC algorithm can be synchronously realized, a system realized by an optical method of the NGRC algorithm is simple in structure, the requirements on a manufacturing process are low, and meanwhile, the system performance requirements are guaranteed.

Description

Optical signal processing device and related method

Technical Field

The present disclosure relates to the field of optical communications technologies, and in particular, to an optical signal processing apparatus and a related method.

Background

Deep learning technology based on deep neural network is widely applied to the hot fields of voice recognition, image processing, medical diagnosis, automatic driving and the like at present. The deep neural network is mainly divided into a feedforward neural network and a circulating neural network. The cyclic neural network has feedback connection from back to front, has memory and association capability, and can realize the processing and prediction of time sequence information in a dynamic system, so that a great deal of researches are carried out. In the cyclic neural network, the reserve pool computing (reservoir computing, RC) model is simple in structure, low in training difficulty and capable of achieving good accuracy. For special application scenarios such as a hardware platform with low configuration, reservoir calculation is a very suitable neural network model.

In recent years, researchers have also proposed next generation pool computing (next generation reservoir computing, NGRC) algorithms. NGRC is a neural network based on nonlinear autoregressive, and has been demonstrated to be equivalent to RC. Compared with the random connection of RC nodes, the feature vector space of the NGRC has an explicit relation, the prediction performance equivalent to the RC with large-scale nodes can be realized only by a small feature vector space, and the multi-step prediction performance under a specific scene is better. The conventional method for realizing NGRC is realized based on integrated circuit chips (such as CPU and GPU), and has the defects of universality in large-scale and high-complexity neural networks realized by the integrated circuits, such as higher process requirements, poorer high-speed calculation stability, high power consumption, low energy consumption ratio and the like.

Disclosure of Invention

The embodiment of the application provides an optical signal processing device and a related method, by arranging a Brillouin optical signal read-write unit, various functions in an NGRC algorithm can be synchronously realized, so that a system realized by an optical method of the NGRC algorithm has a simple structure, has low requirements on a manufacturing process, and simultaneously ensures the performance requirement of the system.

In a first aspect, an optical signal processing apparatus is provided, the apparatus including a brillouin optical signal reading and writing unit:

the Brillouin optical signal reading and writing unit is used for receiving a writing optical signal, a first reading optical signal and a second reading optical signal, wherein the writing optical signal and the second reading optical signal bear a first signal to be processed;

the Brillouin optical signal reading and writing unit is further used for realizing a linear dimension-increasing function according to the writing optical signal and the first reading optical signal to obtain a first output optical signal;

the brillouin optical signal reading and writing unit is further configured to implement a nonlinear mapping function according to the writing optical signal and the second reading optical signal, and obtain a second output optical signal.

In this embodiment of the present application, the brillouin optical signal reading and writing unit receives the writing optical signal, the first reading optical signal and the second reading optical signal, both of which carry the first signal to be processed, and the first reading optical signal does not carry information. The Brillouin optical signal unit obtains a first output optical signal by combining a phonon signal of the writing optical signal after the Brillouin scattering effect and a first reading optical signal to realize a linear dimension increasing function, and realizes a nonlinear mapping function by combining the phonon signal and a second reading optical signal. The process can complete two key functions in the optical method implementation process of the NGRC algorithm through one functional unit. The complexity of system layout is reduced, the requirements on the manufacturing process are reduced, and the system performance is ensured.

In one possible design, the brillouin optical signal read-write unit includes a write optical waveguide, a read optical waveguide, a photoacoustic conversion waveguide, a phonon free propagation region, and an acousto-optic conversion waveguide;

a writing optical waveguide for receiving the writing optical signal and inputting the writing optical signal into the photoacoustic conversion waveguide;

a photoacoustic conversion waveguide for converting the write optical signal into a phonon signal;

a phonon free propagation region for propagating a phonon signal from the photoacoustic conversion waveguide to the acousto-optic conversion waveguide;

a reading optical waveguide for receiving the first reading optical signal and inputting the first reading optical signal into the acousto-optic conversion waveguide;

and the acousto-optic conversion waveguide is used for combining the first read optical signal and the phonon signal to realize a linear dimension-increasing function and obtain a first output optical signal.

In one possible design, the reading optical waveguide is further configured to receive a second reading optical signal and input the second reading optical signal into the acousto-optic conversion waveguide;

the acousto-optic conversion waveguide is further used for combining the second read optical signal and the phonon signal to achieve a nonlinear mapping function and obtain a first output optical signal.

In one possible design, the device further comprises a first laser, a second laser, a modulator, an optical beam splitter, wherein the first laser is connected with the modulator, the modulator is connected with the optical beam splitter, and the optical beam splitter and the second laser are connected with the brillouin optical signal read-write unit;

A first laser for generating a first beam;

the modulator is used for loading a first signal to be processed on the first light beam to obtain a first optical signal;

the optical beam splitter is used for dividing the first optical signal into two paths of optical signals, wherein one path of optical signal is a writing optical signal, and the other path of optical signal is a first reading optical signal;

and the second laser is used for generating a second light beam and taking the second light beam as a second reading optical signal.

In one possible design, the apparatus further comprises a first light combiner; the first optical combiner is connected with the Brillouin optical signal reading and writing unit;

the first beam combiner is configured to receive the first output optical signal and the second output optical signal, and perform linear combination on the first output optical signal and the second output optical signal to obtain a combined optical signal.

In one possible design, the device further comprises a third laser and a second optical combiner connected with the third laser, wherein the second optical combiner is also connected with the optical beam splitter and the brillouin optical signal reading and writing unit;

a third laser for generating a third light beam as a control light signal;

the second optical combiner is used for receiving the control optical signal and the writing optical signal, inputting the control optical signal and the writing optical signal into the Brillouin optical signal reading and writing unit, and the control optical signal is used for providing energy for converting the writing optical signal into a phonon signal.

In one possible design, the first optical combiner is also connected to a detector, which is connected to the computing device;

the detector is used for detecting the combined optical signal and converting the combined optical signal into a combined electrical signal;

the computing device is used for obtaining the combined electrical signal, obtaining an NGRC characteristic space according to the combined electrical signal, multiplying the NGRC characteristic space by the NGRC weight matrix, and obtaining an NGRC prediction result.

In the embodiment of the application, a complete optical signal processing device is obtained through the combined connection of a series of components, the device can completely perform the optical method implementation process of the NGRC algorithm, the structure is simple, the device manufacturing process requirement is low, the processing efficiency is high, and the system performance of the optical method implementation of the NGRC algorithm is ensured. Furthermore, the device can realize customization of NGRC algorithm by adjusting the length of the phonon free propagation region, the combination of the writing optical signal and the reading optical signal, thereby being applicable to different application scenes and practical problems.

In a second aspect, there is provided an optical signal processing system comprising N parallel optical signal processing devices as in the first aspect or any one of the first aspects.

In one possible design, the system includes a brillouin optical signal read-write array, which is composed of N brillouin optical signal read-write units.

In one possible design, the system further includes N first lasers and N modulators connected to the N first lasers, where the N first lasers are used to generate N first light beams, and the N modulators are used to load first signals to be processed for the N first light beams, respectively, so as to obtain N first optical signals;

the system further includes a second laser for generating a second beam of light, the second beam of light being the second read optical signal;

the N first optical signals and the second read optical signals are combined into one optical signal through a first optical beam combiner, one optical signal is divided into two optical signals through a first optical beam splitter, a first optical signal in the two optical signals passes through an array waveguide grating, the array waveguide grating is used for dividing the first optical signal into N write optical signals according to data signals, a second optical signal in the two optical signals is divided into N first read optical signals corresponding to the data signals through a second optical beam splitter, and N second read optical signals;

N paths of writing optical signals, N paths of first reading optical signals and N paths of second reading optical signals are respectively input into N Brillouin optical signal reading and writing units, wherein one path of writing optical signals input by each Brillouin optical signal reading and writing unit corresponds to one path of data signals of the first reading optical signals.

In one possible design, the system further comprises a third laser, and a third optical splitter connected to the third laser, the third optical splitter further connected to N second optical combiners, each of the N second optical combiners being connected to one brillouin optical signal reading and writing unit,

the third laser is used for generating a third light beam as a control light signal, the third light beam splitter receives the control light signal and divides the control light signal into N paths of control light signals, and the N paths of control light signals are respectively input into N second light beam combiners;

each of the N second optical combiners receives one of the N writing optical signals respectively, and the writing optical signals are input to a Brillouin optical signal reading and writing unit connected with a control optical signal, and the control optical signal is used for providing energy for converting the writing optical signals into phonon signals.

In one possible design, when the second optical beam splitter splits the control optical signal into N control optical signals, the control optical signal splitting ratio is determined according to the weights of the N brillouin optical signal reading and writing units.

In the embodiment of the application, the optical method implementation of the NGRC algorithm is performed through the parallel architecture of the wavelength division multiplexing technology, so that the processing of the multidimensional data signal can be completed, and the optical method implementation of the NGRC algorithm of the multidimensional signal can be completed. The core of the system is that a plurality of Brillouin optical signal read-write units are arranged in parallel, the connection mode is simple, the complexity of the manufacturing process is low, and the high-efficiency realization of the NGRC algorithm is achieved. In addition, the parallel architecture can improve the per-unit area calculation force, is highly customizable, and is suitable for the data signal processing problem of different dimensions.

In a third aspect, there is provided an information acquisition method applied to an optical signal processing apparatus including a brillouin optical signal reading and writing unit, the method comprising:

receiving a writing optical signal, a first reading optical signal and a second reading optical signal through a Brillouin optical signal reading and writing unit, wherein the writing optical signal and the second reading optical signal bear a first signal to be processed;

the Brillouin optical signal reading and writing unit realizes a linear dimension increasing function according to the writing optical signal and the first reading optical signal to obtain a first output optical signal;

the Brillouin optical signal reading and writing unit realizes a nonlinear mapping function according to the writing optical signal and the second reading optical signal, and obtains a second output optical signal.

In one possible design, the brillouin optical signal read-write unit includes a write optical waveguide, a read optical waveguide, a photoacoustic conversion waveguide, a phonon free propagation region, and an acousto-optic conversion waveguide; the method further comprises the steps of:

receiving a writing optical signal through a writing optical waveguide, and inputting the writing optical signal into a photoacoustic conversion waveguide;

converting the writing optical signal into a phonon signal through a photoacoustic conversion waveguide;

propagating phonon signals from the photoacoustic conversion waveguide to the acousto-optic conversion waveguide through the phonon free propagation region;

receiving a first reading optical signal through a reading optical waveguide, and inputting the first reading optical signal into an acousto-optic conversion waveguide;

and realizing a linear dimension-increasing function by combining the first read optical signal and the phonon signal through the acousto-optic conversion waveguide, and obtaining a first output optical signal.

In one possible design, the method further comprises: receiving a second reading optical signal through the reading optical waveguide, and inputting the second reading optical signal into the acousto-optic conversion waveguide; and realizing a nonlinear mapping function by combining the acousto-optic conversion waveguide with the second read optical signal and the phonon signal, and obtaining a first output optical signal.

In one possible design, the device further comprises a first laser, a second laser, a modulator and an optical beam splitter, wherein the first laser is connected with the modulator, the modulator is connected with the optical beam splitter, and the optical beam splitter and the second laser are connected with the brillouin optical signal reading and writing unit; the method further comprises the steps of: generating a first beam by a first laser; loading a first signal to be processed on the first light beam through a modulator to obtain a first optical signal; dividing the first optical signal into two paths of optical signals through an optical beam splitter, wherein one path of optical signal is a writing optical signal, and the other path of optical signal is a first reading optical signal; a second beam is generated by a second laser, and the second beam is used as a second reading optical signal.

In one possible design, the apparatus further comprises a first light combiner; the first optical combiner is connected with the Brillouin optical signal reading and writing unit; the method further comprises the steps of: and receiving the first output optical signal and the second output optical signal through the first optical combiner, and linearly combining the first output optical signal and the second output optical signal to obtain a combined optical signal.

In one possible design, the device further comprises a third laser and a second optical combiner connected with the third laser, wherein the second optical combiner is also connected with the optical beam splitter and the brillouin optical signal reading and writing unit; the method further comprises the steps of: generating a third light beam as a control light signal by a third laser; and receiving a control optical signal and a writing optical signal through a second optical combiner, and inputting the control optical signal and the writing optical signal into a Brillouin optical signal reading and writing unit, wherein the control optical signal is used for providing energy for converting the writing optical signal into a phonon signal.

In one possible design, the first optical combiner is also connected to a detector, which is connected to the computing device; the method further comprises the steps of: detecting the combined optical signal by a detector and converting the combined optical signal into a combined electrical signal; and acquiring the combined electric signal through a computing device, acquiring an NGRC characteristic space according to the combined electric signal, and multiplying the NGRC characteristic space by an NGRC weight matrix to acquire an NGRC prediction result.

In a fourth aspect, there is provided an optical signal processing method comprising steps for performing the functions of the system of the second aspect or any one of the second aspects, for example.

In a fifth aspect, there is provided a communication apparatus comprising:

a memory for storing instructions; and

at least one processor coupled to the memory;

wherein the instructions, when executed by the at least one processor, cause the processor to perform the method of any of the third or fourth aspects.

In a sixth aspect, embodiments of the present application provide a chip system, including: a processor coupled to a memory for storing a program or instructions that when executed by the processor cause the chip system to implement the method of any one of the third aspect or the third aspect, or cause the chip system to implement the method of the fourth aspect.

Optionally, the system on a chip further comprises an interface circuit for interacting code instructions to the processor.

Alternatively, the processor in the chip system may be one or more, and the processor may be implemented by hardware or software. When implemented in hardware, the processor may be a logic circuit, an integrated circuit, or the like. When implemented in software, the processor may be a general purpose processor, implemented by reading software code stored in a memory.

Alternatively, the memory in the system-on-chip may be one or more. The memory may be integral with the processor or separate from the processor, and is not limited in this application. For example, the memory may be a non-transitory processor, such as a ROM, which may be integrated on the same chip as the processor, or may be separately provided on different chips, and the type of memory and the manner of providing the memory and the processor are not specifically limited in this application.

In a seventh aspect, embodiments of the present application provide a computer-readable storage medium having stored thereon a computer program or instructions that, when executed, cause a computer to perform the method of any one of the above third aspect or the third aspect, or cause a computer to perform the method of the above fourth aspect.

In an eighth aspect, embodiments of the present application provide a computer program product which, when read and executed by a computer, causes the computer to perform the method in any one of the possible implementations of the third aspect or the third aspect, or causes the computer to perform the method of the fourth aspect.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments will be briefly described below.

FIG. 1 is a schematic diagram of an RC algorithm according to an embodiment of the present disclosure;

FIG. 2 is a parallel architecture of an RC system on optical chips according to an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of an RC system on optical chips with serial architecture according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of an NGRC algorithm according to an embodiment of the present disclosure;

FIG. 5 is an optical implementation system architecture of the NGRC algorithm provided in embodiments of the present application;

fig. 6A is a schematic diagram of an optical signal processing device according to an embodiment of the present application;

fig. 6B is a schematic diagram of a brillouin optical signal read-write unit according to an embodiment of the present application;

fig. 6C is a timing diagram of implementing a linear dimension increasing function of the brillouin optical signal read-write unit according to the embodiment of the present application;

fig. 6D is a schematic structural diagram of another optical signal processing apparatus according to an embodiment of the present disclosure;

FIG. 7A is a schematic diagram of another optical signal processing system according to an embodiment of the present disclosure;

FIG. 7B is a schematic diagram of an optical NGRC simulation Lorenz63 chaotic system according to an embodiment of the present disclosure;

Fig. 8 is a schematic diagram of another optical signal processing apparatus according to an embodiment of the present disclosure;

fig. 9 is a schematic diagram of an optical signal processing method according to an embodiment of the present application;

fig. 10 is a schematic hardware structure of a communication device in an embodiment of the present application.

Detailed Description

The terms "first," "second," "third," and "fourth" and the like in the description and in the claims of this application and in the drawings, are used for distinguishing between different objects and not for describing a particular sequential order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

"plurality" means two or more. "and/or", describes an association relationship of an association object, and indicates that there may be three relationships, for example, a and/or B, and may indicate: a exists alone, A and B exist together, and B exists alone. The character "/" generally indicates that the context-dependent object is an "or" relationship.

The prior art will be first described with reference to the drawings.

First, a related art of the RC algorithm will be described.

Referring to fig. 1, fig. 1 is a schematic diagram of an RC algorithm according to an embodiment of the disclosure. The RC algorithm consists of an input layer, a reserve tank and an output layer. Wherein the input layer and the output layer correspond to the input signal and the output signal, respectively. The reserve pool is a node network formed by a plurality of nodes, and the topological connection W (internal weight matrix) of each node and other nodes is random connection inside the reserve pool. The state of each node in the reserve pool is determined by the node state at the previous moment and the input value at the current moment, wherein the node state at the previous moment is multiplied by a weight matrix W in the reserve pool to obtain information of historical moment, and the input layer is multiplied by the input weight matrix W _in And obtaining the information of the current moment. Input weight matrix W _in Is randomly generated. As the input signal increases, the internal state of the reservoir is also constantly updated. Multiplying the pool state by the output weight matrix W _out A predicted output signal may be obtained. Here, a weight matrix W is output _out Is obtained by model training based on a training set.

When implementing the RC algorithm by an electrical method, the main drawbacks are that the internal connection is random connection, electrical feedback is required, and a large node scale is required to achieve higher calculation accuracy.

In the prior art, an RC algorithm can be realized by an optical method, and the RC algorithm specifically comprises a serial architecture and a parallel architecture. Referring to fig. 2, fig. 2 is an optical RC system with parallel architecture according to an embodiment of the present application. As shown in fig. 2, the operating principle of the RC system is that a node network composed of multiple nodes circulates optical energy (information). The RC system contains core devices including low loss waveguides, beam splitters (not shown), beam combiners (not shown), and optical delay lines. The optical delay line is used for delaying the optical signals, so that the optical signals at different moments can be obtained conveniently. The optical waveguide is used for transmission of optical signals between nodes. The parallel structure can process the optical information in parallel, so that the unit area calculation force and the node number are higher, and the parallel structure can be combined with various multiplexing modes.

Referring to fig. 3, fig. 3 is an RC system on an optical chip with a serial architecture according to an embodiment of the present application. As shown in fig. 3, the operating principle of the RC system is that a loop structure formed by single nodes circulates light energy (information). The RC system comprises core devices including optical delay lines, photodetectors (readout photodiodes in the figure), and electrical amplifiers. The optical signals are generated by a laser, the electric signals are generated by an arbitrary waveform generator (arbitrary waveform generator, AWG), the optical signals are modulated by Mach-Zehnder Modulator, M-Z and then are used as input signals of the storage pool to enter the storage pool, and a part of the optical signals in the storage pool are converted into electric signals by a reading photodiode and are further processed by a computer. The other part of the optical signals are converted into feedback electric signals through an optical attenuator and a feedback photodiode, the feedback electric signals and the electric signals to be processed are overlapped through a mixer, electric power amplification is carried out through an electric amplifier, and then the feedback electric signals and the electric signals are input into an M-Z as electric modulation signals. Wherein the optical attenuator is used to adjust the feedback weight,representing the input weights for adjustment. The serial structure can reduce the size of the device, and reduce the process requirement and the power consumption of the device.

The optical implementation of the RC algorithm has its corresponding drawbacks. For example, in a parallel architecture, as the node scale in the RC network increases, the system power consumption increases, the calculation accuracy decreases, and meanwhile, the device manufacturing process has higher requirements. The serial structure is essentially that a plurality of virtual nodes are obtained for executing operation on a single element in a time division multiplexing way, the processing speed and the parallelism are relatively poor, the number of the virtual nodes is limited by time delay, bandwidth and electrical response, and the wavelength division, the mode division and the polarization multiplexing can increase the complexity of the system. However, whether the serial architecture or the parallel architecture is adopted, the RC algorithm needs to feed back the state of the reserve pool at the last moment, and an optical feedback path or an electrical feedback path is needed in physical implementation, so that the implementation of all-optical feedback is difficult, the electrical feedback speed is slow, and the running speed of the whole system can be limited.

The NGRC algorithm differs from the RC algorithm. FIG. 4 is a schematic diagram of an NGRC algorithm according to an embodiment of the present application, as shown in FIG. 4, for processing three-dimensional signals _i ,i,i]For example, where the subscript i indicates the different times. Firstly, signal values at different moments are linearly combined to obtain a linear feature space, and the result in the figure is the linear feature space formed by three-dimensional signal values at the current moment i and the previous moment i-1. And multiplying the elements in the linear characteristic space at different moments by one to obtain a nonlinear characteristic space. The nonlinear feature space constructed by multiplying the signal value of each dimension at each current instant i by the signal value of each dimension at the previous instant i-1 is shown. Two functions are realized by this process of constructing a nonlinear feature space: 1. generating a nonlinear component; 2. the signal values at different times are linked. Then, the linear feature space and the nonlinear feature space are spliced to obtain an NGRC feature space, and the mathematical form of the NGRC feature space is a one-dimensional vector. Finally, the weight matrix W is output ^out Multiplying the NGRC characteristic space to obtain the signal predictive value of i+1 at the next moment. Wherein W is ^out Model training based on the training set is required.

Existing methods implement NGRC algorithms based on integrated circuit chips such as central processing units (central processing unit, CPU), graphics processing units (graphics processing unit, GPU). But there are common disadvantages of implementing a neural network such as large scale and high complexity by an integrated circuit, such as high process requirements, poor high-speed calculation stability, large power consumption, low energy consumption, etc. Based on this we propose an optical implementation of the NGRC algorithm. And the optical method implementation system architecture is shown in fig. 5, and fig. 5 is an optical implementation system architecture of an NGRC algorithm provided in the embodiment of the present application. The process mainly comprises six parts of signal loading, linear dimension lifting, nonlinear mapping, linear combination, photoelectric detection and calculation equipment processing:

1. and (3) signal loading: the function of signal loading is to load the electrical signal to be processed onto the input optical carrier. The system supports multi-dimensional modulation modes such as wavelength division multiplexing, mode division multiplexing and the like, so that an input signal can be a multi-wavelength optical signal, a multi-mode optical signal and the like.

2. Linear dimension increase: by combining the optical signals at different moments, the purpose of linear dimension increase of the optical signals is achieved, so that the optical signals representing the NGRC linear characteristic space are generated. The method specifically corresponds to the process of linearly combining signal values at different moments to obtain a linear characteristic space.

3. Nonlinear mapping: and combining and mapping the NGRC linear characteristic space optical signals after the dimension rise according to a specific relation to obtain the optical signals representing the NGRC nonlinear characteristic space. The method specifically corresponds to the process of multiplying elements in the linear characteristic space at different moments by each other to obtain the nonlinear characteristic space.

4. Linear combination: and combining the optical signals of the NGRC linear characteristic space and the NGRC nonlinear characteristic space to obtain an optical signal representing the NGRC total characteristic space.

5. Photoelectric detection: the NGRC total feature space is converted from an optical signal to an electrical signal for further processing by the computing device.

6. Computing device processing: the computing device is primarily implemented by a general purpose computer, a field programmable gate array (field programmable gate array, FPGA) or an application specific integrated circuit (application specific integrated circuit, ASIC). The functions performed by the computing device include: (1) Training the NGRC model based on the training set to obtain an output weight matrix; (2) transmitting the electrical signal to be processed to a signal loading module; (3) And multiplying the output NGRC electric signal with the output weight matrix to obtain a final output result.

In the above process, the system architecture realized by the optical method can be deduced based on the principle of the NGRC algorithm, but the specific device selection, connection mode and realization process are not determined. And because of the difference of the implementation principle of the RC algorithm and the NGRC algorithm and the defects of the RC system on the optical sheet, the specific mode for determining the optical method implementation of the NGRC algorithm can not be realized according to the optical method of the RC algorithm.

Referring to fig. 6A, a schematic diagram of an optical signal processing device is provided in an embodiment of the present application. As shown in fig. 6A, the apparatus includes a brillouin optical signal reading and writing unit, and the brillouin optical signal reading and writing unit is specifically configured to implement the following functions:

201. the Brillouin optical signal reading and writing unit is used for receiving a writing optical signal, a first reading optical signal and a second reading optical signal, wherein the writing optical signal and the second reading optical signal bear a first signal to be processed.

202. The brillouin optical signal reading and writing unit is further used for realizing a linear dimension-increasing function according to the writing optical signal and the first reading optical signal, and obtaining a first output optical signal.

203. The brillouin optical signal reading and writing unit is further configured to implement a nonlinear mapping function according to the writing optical signal and the second reading optical signal, and obtain a second output optical signal.

The brillouin effect refers to a process of achieving the mutual conversion of an optical signal and an acoustical sub-signal. The brillouin optical signal read-write unit in the embodiment of the application can realize a linear dimension-increasing process and a nonlinear mapping process in the process by the optical method of the NGRC algorithm through the mutual conversion of the optical signal and the phonon signal. As shown in fig. 6A, the brillouin optical signal reading and writing unit receives the writing optical signal, the first reading optical signal, and the first reading optical signal. The write optical signal carries a first signal to be processed, for example, the first signal to be processed may be the signal of the time i described above. The first read optical signal does not bear signals, but the write optical signal is converted into phonon signals from the optical signals after being input into the Brillouin optical signal read-write unit, and the propagation speed of the phonon signals is slower than that of the optical signals, so that the signal at the moment i generates time delay to obtain the signal at the moment i+1, namely the first output optical signal, and the linear dimension increasing function is realized. The second read optical signal carries a first signal to be processed, namely a signal of i, and is subjected to element multiplication with a phonon signal output by the Brillouin optical signal, namely a signal of i+1, so that a nonlinear characteristic space is obtained, namely a second output optical signal, and a nonlinear mapping function is realized.

Referring to fig. 6B, fig. 6B is a block diagram of a brillouin optical signal read-write unit provided in the embodiment of the present application, and as shown in fig. 6B, the brillouin optical signal read-write unit 10 may specifically include a write optical waveguide 11, a read optical waveguide 12, a photoacoustic conversion waveguide 13, a phonon free propagation region 14, and an acousto-optic conversion waveguide 15, where these components specifically implement a linear dimension increasing function and a nonlinear mapping function in the optical implementation method of the NGRC algorithm through the following steps:

s1, a writing optical waveguide is used for receiving the writing optical signal and inputting the writing optical signal into the photoacoustic conversion waveguide.

Referring to fig. 6C, a timing diagram for implementing a linear dimension-increasing function of a brillouin optical signal read-write unit according to an embodiment of the present application, as shown in fig. 6C, a write optical signal carries a signal to be processed, which is a signal at three moments of a (t-1), a (t), and a (t+1).

S2, a photoacoustic conversion waveguide is used for converting the writing optical signal into a phonon signal.

The writing optical signal is converted into a phonon signal by the Brillouin scattering effect in the photoacoustic conversion waveguide, the generation time of the phonon signal is the writing end time of the writing optical signal, and the phonon signal end time is the reading time of the reading optical signal (including the first reading optical signal or the second reading optical signal). As shown in fig. 6C, the phonon signal at the time a (t-1) is generated after the optical signal at the time a (t-1) is written (including when the optical signal at the time a (t) is generated), and the phonon signal at the time a (t) is generated after the optical signal at the time a (t) is written (including when the optical signal at the time a (t+1) is generated). And so on.

And S3, a phonon free propagation region for propagating phonon signals from the photoacoustic conversion waveguide to the acousto-optic conversion waveguide.

The phonon signal can be propagated in a phonon free propagation region, and then propagated from the photoacoustic conversion waveguide to the acousto-optic conversion waveguide, and then converted into an optical signal by the acousto-optic conversion waveguide. The Brillouin scattering effect in the process causes the signal to undergo a series of conversion from writing optical signal to phonon signal to optical signal, generates signal delay, and obtains a signal at a (t-1) corresponding to a (t) time and a signal at a (t) corresponding to a (t+1) time.

The length of the phonon free propagation region determines the signal delay time, and the length of the phonon free propagation region is proportional to the signal delay time. Then, the length of the phonon free propagation region can be customized according to the signal delay requirement so as to meet the optical signal delay of different scenes and different requirements.

And S4, a reading optical waveguide is used for receiving the first reading optical signal and inputting the first reading optical signal into the acousto-optic conversion waveguide.

The acousto-optic conversion waveguide includes an optical signal after the phonon signal is converted, and the first read optical signal and the converted optical signal may be combined in the acousto-optic conversion waveguide.

S5, the acousto-optic conversion waveguide is further used for combining the first read optical signal and the phonon signal to achieve a linear dimension increasing function, and a first output optical signal is obtained.

The first read optical signal remains synchronized with the write optical signal, but the write optical signal undergoes a conversion by the brillouin effect, which results in a delay in the signal. That is, the combination of the first read optical signal and the phonon signal, that is, the combination of the write optical signal at the a (t) time and the first read optical signal at the a (t) time is realized at the a (t) time, the combination of the write optical signal at the a (t) time and the first read optical signal at the a (t+1) time is realized at the a (t+1) time, and the first output optical signal has both the optical signal at the current time and the optical signal at the next time at the same time, thereby realizing the linear dimension increasing function. Since the first read optical signal does not carry information, the combination of a (t-1) x 1 is implemented at time (t), and accordingly, the combination of a (t) x 1 is implemented at time a (t+1). Resulting in a first output optical signal as shown in fig. 6C.

Further, the device can also realize a nonlinear mapping function, and specifically comprises the following steps:

s6, a reading optical waveguide is used for receiving the second reading optical signal and inputting the second reading optical signal into the acousto-optic conversion waveguide.

And S7, an acousto-optic conversion waveguide is used for combining the second read optical signal and the phonon signal to realize a nonlinear mapping function, so as to obtain a second output optical signal.

The second read optical signal carries signals at three moments of a (t-1), a (t) and a (t+1), and the acousto-optic conversion waveguide comprises a delayed optical signal. Then the second read optical signal is combined with the optical signal after phonon signal conversion, so that the linear characteristic of a (t) a (t-1) can be obtained at a (t) time, the linear characteristic of a (t+1) a (t) is obtained at a (t+1) time, and so on, so as to obtain a second output optical signal as shown in fig. 6C, and the nonlinear mapping function is realized.

It can be seen that, in the embodiment of the present application, the writing optical signal, the first reading optical signal and the second reading optical signal are received by the brillouin optical signal reading and writing unit, the writing optical signal and the second reading optical signal both bear the first signal to be processed, and the first reading optical signal does not bear information. The Brillouin optical signal unit obtains a first output optical signal by combining a phonon signal of the writing optical signal after the Brillouin scattering effect and a first reading optical signal to realize a linear dimension increasing function, and realizes a nonlinear mapping function by combining the phonon signal and a second reading optical signal. The process can complete two key functions in the optical method implementation process of the NGRC algorithm through one functional unit. The complexity of system layout is reduced, the requirements on the manufacturing process are reduced, and the system performance is ensured. Furthermore, the device can realize customization of NGRC algorithm by adjusting the length of the phonon free propagation region, the combination of the writing optical signal and the reading optical signal, thereby being applicable to different application scenes and practical problems.

Alternatively, referring to fig. 6D, fig. 6D is a schematic structural diagram of another optical signal processing apparatus according to an embodiment of the present application. As shown in fig. 6D, the apparatus may further include a first laser 20, a second laser 21, a modulator 30, and an optical beam splitter 40. The first laser 20 is used for generating a first light beam, and is connected to a modulator. The modulator is configured to load a first signal to be processed on the first light beam to obtain a first optical signal when the first light beam propagates to the modulator. The other end of the modulator is also connected with an optical beam splitter, and when the first optical signal is transmitted to the optical splitter, the first optical signal is split into two paths of optical signals, wherein one path of optical signal is a writing optical signal, and the other path of optical signal is a first reading optical signal. Therefore, the write optical signal carries the same first signal to be processed as the first read optical signal. The other end of the optical splitter is connected to the brillouin optical signal reading and writing unit, and specifically, the writing optical waveguide described in the foregoing embodiment may be connected to the photoacoustic conversion waveguide of the brillouin optical signal, and the writing optical signal and the first reading optical signal may be propagated to the acousto-optic conversion waveguide through the writing optical waveguide to perform acousto-optic conversion.

The second laser 21 may be used to generate a second light beam, and the second light beam may be directly input to the brillouin optical signal reading and writing unit as a second read optical signal. Alternatively, the second read optical signal and the first read optical signal may be combined and input to the brillouin optical signal unit by the third optical combiner 52, and the linear dimension increasing function and the nonlinear mapping function may be realized.

Optionally, the optical signal processing apparatus may further include a first optical combiner 50, where the first optical combiner 50 is connected to the brillouin optical signal read-write unit 10, and is configured to receive the first output optical signal generated in the linear dimension-increasing process and the second output optical signal generated in the nonlinear mapping process, and perform linear combination on the first output optical signal and the second output optical signal to obtain a combined optical signal.

Alternatively, the optical signal processing device may comprise a third laser 22 and a second optical combiner 51 connected to each other. Wherein the third laser 22 is arranged to generate a third light beam as the control light signal. The control optical signal does not carry data information (as shown in fig. 6C, described above), but may provide energy to convert the write optical signal into the phonon signal. Thus, after the third laser 22 generates the control optical signal, the control optical signal is combined with the write optical signal by the second optical combiner 51. The other end of the second optical combiner 51 is connected to the brillouin optical signal reading/writing unit 10, and receives the control optical signal and the writing optical signal output from the second optical combiner 51. Specifically, as shown in fig. 6B described above, the control optical signal and the write optical signal may be transmitted to the brillouin optical signal unit through the write optical waveguide. The control optical signal is used for providing energy, so that the control optical signal can also play a role of turning on and off the brillouin optical signal read-write unit, i.e. in case the control optical signal is provided sufficiently, the brillouin optical signal read-write unit is turned on, otherwise the brillouin optical signal read-write unit is turned off.

Further, the optical signal processing apparatus may further include a detector 60 and a computing device 70 connected to each other, wherein the other end of the detector 60 is connected to the first optical combiner 50 for detecting the combined optical signal and converting the combined optical signal into a combined electrical signal; the computing device 70 is configured to obtain a combined electrical signal, obtain an NGRC feature space according to the combined electrical signal, multiply the NGRC feature space with an NGRC weight matrix, and obtain an NGRC prediction result. Therefore, the optical signal processing device provided by the embodiment of the application completes the whole process of realizing the NGRC algorithm function shown in fig. 5 through the arrangement and connection of each component.

It should be noted that, assuming that the prediction result at the i+1 time is output by the computing device, the prediction result at the i+1 time may be used to predict the result at the i+2 time, so that the prediction result output by the computing device 70 may be used as the input result of the modulator 30 to perform the assignment of the signal to be processed in the next round, to form an iterative NGRC algorithm execution process.

Therefore, in the embodiment of the application, a complete optical signal processing device is obtained through the combined connection of a series of components, the device can completely perform the optical method implementation process of the NGRC algorithm, the structure is simple, the device manufacturing process requirement is low, the processing efficiency is high, and the system performance realized by the optical method of the NGRC algorithm is ensured.

The embodiment of the application also provides a parallel architecture of another wavelength division multiplexing technology for implementing the optical method of the NGRC algorithm. That is, the embodiment of the present application also protects an optical signal processing system, including N brillouin optical signal processing apparatuses described in any one of fig. 6A to 6D.

Specifically, referring to fig. 7A, fig. 7A is a schematic diagram of another optical signal processing system provided in an embodiment of the present application, and as shown in fig. 7A, the optical signal processing system includes a brillouin optical signal read-write array. Consists of N Brillouin optical signal read-write units (N=3 in the figure). Each brillouin optical signal read-write unit can complete a group of linear ascending and/or nonlinear mapping functions, and N brillouin optical signal read-write units can complete the linear ascending and/or nonlinear mapping functions of N-dimensional parameters.

Optionally, the system may further comprise N first lasers, each connected to a modulator for loading the generated laser signal with N-dimensional data signals. The embodiment of the present application is described by taking n=3 as an example. As shown in fig. 7A, the N first lasers are a first laser 201, a first laser 202, and a first laser 203, respectively. Each of the three first lasers is connected to a modulator, wherein the first laser 201 is connected to a modulator 211, the first laser 202 is connected to a modulator 212, and the first laser 203 is connected to a modulator 213. Each of the three lasers is configured to generate a first light beam having a different wavelength from the other lasers, and a modulator coupled to each of the first lasers is configured to load a first signal to be processed (three-dimensional data signal) for each of the first light beams to obtain three first optical signals. A second laser 200 is also included for generating a second beam. The second light beam is directly used as a second reading optical signal without loading a signal to be processed on the second light beam.

The three first optical signals and the second read optical signal form one optical signal through the first optical combiner 221. The first optical beam splitter 231 then splits the one optical signal into two optical signals. The first optical signal of the two optical signals is input into the arrayed waveguide grating 240, the first optical signal is separated by the arrayed waveguide grating 240 according to three paths of data signals, and each path of data signal is input into a brillouin optical signal reading and writing unit as a writing optical signal. The second optical signal divided by the first optical splitter 231 includes four laser beams, that is, three first optical signals carrying different data signals and a second read optical signal not carrying a data signal. The second optical signal is input to the second optical splitter 232 and is divided into three sub-signals by the second optical splitter 232, and in the three sub-signals, the content corresponding to the three data signals in each sub-signal may be referred to as a first read optical signal and the content corresponding to the second optical beam without the data signal may be referred to as a second read optical signal. Three signals are input to three brillouin optical signal reading and writing units 251, 252, and 253, respectively, for performing a linear dimension-up function and a nonlinear mapping function in combination with a phonon signal obtained by delaying a writing optical signal (the phonon signal has been converted into an optical signal at the time of combination). Wherein the first optical signal and the second optical signal remain synchronized. In this way, the linear dimension increasing function and the nonlinear mapping function can be realized simultaneously.

Optionally, the optical signal processing system further comprises a third laser 204, and a third optical splitter 233 connected to the third laser 204. Wherein the third laser 204 is configured to generate a third light beam as a control light signal, and the third beam splitter 233 receives the control light signal and splits the control light signal into three control light signals. Consistent with the foregoing description, the control optical signal is used to supply energy to the writing optical signal, and further may be used to control the opening and closing of the brillouin optical signal reading and writing unit. Therefore, each path of control optical signal and each path of write optical signal generated by the above-mentioned generation are combined by a second optical combiner (second optical combiners 222, 223 and 224 in the figure respectively) and are respectively input into a brillouin optical signal read-write unit.

In an alternative case, when the second optical beam splitter splits the control optical signal into three paths of control optical signals, the control optical signal splitting ratio may be determined according to weights of the three connected brillouin optical signal reading and writing units. That is, the control optical signal dividing ratio can be regulated and controlled according to the weight change of the brillouin optical signal reading and writing unit.

In addition, as in the foregoing embodiment, the optical signal processing system may further include third optical combiners 225, 226 and 227 connected to the output end of each brillouin optical signal read-write unit, for linearly combining the linear feature space obtained through the linear dimension increasing function and the nonlinear feature space obtained through the nonlinear mapping function in the brillouin optical signal read-write unit, to obtain a complete feature space. The third beam combiner is connected with a detector, and the detector is used for completing a photoelectric detection function and converting optical signals into electric signals. The computing equipment is connected with the detector, receives the electric signals, and interacts with the output weight matrix obtained through training to obtain a final output result. Therefore, the optical signal processing system provided by the embodiment of the application completes the whole process of realizing the NGRC algorithm function shown in fig. 5 through the arrangement and connection of each component.

It can be seen that in the embodiment of the present application, the implementation of the optical method of the NGRC algorithm is performed by using the parallel architecture of the wavelength division multiplexing technology, so that the processing of the multidimensional data signal can be completed, and the implementation of the optical method of the NGRC algorithm of the multidimensional signal can be completed. The core of the system is that a plurality of Brillouin optical signal read-write units are arranged in parallel, the connection mode is simple, the complexity of the manufacturing process is low, and the high-efficiency realization of the NGRC algorithm is achieved. In addition, the parallel architecture can improve the per-unit area calculation force, is highly customizable, and is suitable for the data signal processing problem of different dimensions.

The NGRC algorithm optical implementation system provided by the embodiment of the application has a good prediction effect through simulation verification. Referring specifically to fig. 7B, fig. 7B is a schematic diagram of the result of the optical NGRC simulation Lorenz63 chaotic system according to the embodiment of the present application, and as shown in fig. 7B, lorenz63 is a three-dimensional problem. The simulation uses 3 parallel nodes and 5 different light wavelengths, and generates NGRC linear characteristic space with the size of 12 and nonlinear characteristic space with the size of 27, and the training set length is 47. The simulation process corresponds to sub-graph a) to sub-graph h) for a total of 8 sub-graphs. Wherein, sub-graph a) represents the data distribution of the real data used as the training data set, and sub-graph e) represents the data distribution of the real data used as the test data set. Subgraph b) to subgraph d) represent training results for each dimension in the three-dimensional data of Lorenz63 at a time between lyapunov times 0 to 45 (the abscissa represents lyapunov time). Correspondingly, sub-graph f) through sub-graph h) represent the results of reasoning (predictions) for each dimension in the three-dimensional data of Lorenz63, between Lyapunov times 45-66. The simulation result shows that the standard root mean square error (normalized root mean square error, NRMSE) is 0.06, and the multi-step prediction of about 15 Lyapunov times can be realized, which shows that the optical NGRC of the parallel architecture has higher prediction precision and multi-step prediction capability.

The embodiment of the application also provides other optical signal processing devices for implementing the optical method of the NGRC algorithm. Referring specifically to fig. 8, fig. 8 is a schematic diagram of another optical signal processing apparatus provided in an embodiment of the present application, where, as shown in fig. 8, an optical delay line is used to implement optical signal delay, so as to implement a linear dimension increasing function in an NGRC algorithm. The optical delay line has a simple structure, but is usually large in size, and the optical delay amount which can be realized is limited, so that the optical delay line is more suitable for the situations of less delay amount and single type of delay amount. The device also comprises a nonlinear waveguide for realizing the nonlinear mapping function in the NGRC algorithm. Nonlinear mapping can be generally achieved using a four-wave mixing effect. Also, the optical signal processing device may further include a laser for emitting an optical signal, a modulator for loading a signal to be processed, an optical combiner for combining the optical signal and dividing the optical signal, a detector for performing photoelectric signal conversion, a computing device for processing an electrical signal, and the like.

Compared with the embodiments of fig. 6A to 6D and fig. 7A, the difference between the embodiment and the embodiment of fig. 6A is that the embodiment adopts different devices to respectively realize the linear dimension-increasing and nonlinear mapping functions, and the embodiment has the advantages that the structure customization degree is higher, so that more diversified NGRC characteristic spaces can be realized, and the embodiment has important significance for improving the accuracy and the robustness of the algorithm. The foregoing embodiment has the advantage of simpler system configuration, less need for a number of photodetectors, and the like.

Fig. 9 is a schematic diagram of an optical signal processing method provided in an embodiment of the present application, where the method is applied to an optical signal processing device corresponding to fig. 6A to 6D, and the optical signal processing device includes a brillouin optical signal reading and writing unit, and the method includes the following steps:

301. receiving a writing optical signal, a first reading optical signal and a second reading optical signal through a Brillouin optical signal reading and writing unit, wherein the writing optical signal and the second reading optical signal bear a first signal to be processed;

302. the Brillouin optical signal reading and writing unit realizes a linear dimension increasing function according to the writing optical signal and the first reading optical signal to obtain a first output optical signal;

303. the Brillouin optical signal reading and writing unit realizes a nonlinear mapping function according to the writing optical signal and the second reading optical signal, and obtains a second output optical signal.

It can be seen that, in the embodiment of the present application, the writing optical signal, the first reading optical signal and the second reading optical signal are received by the brillouin optical signal reading and writing unit, the writing optical signal and the second reading optical signal both bear the first signal to be processed, and the first reading optical signal does not bear information. The Brillouin optical signal unit obtains a first output optical signal by combining a phonon signal of the writing optical signal after the Brillouin scattering effect and a first reading optical signal to realize a linear dimension increasing function, and realizes a nonlinear mapping function by combining the phonon signal and a second reading optical signal. The process can complete two key functions in the optical method implementation process of the NGRC algorithm through one functional unit. The complexity of system layout is reduced, the requirements on the manufacturing process are reduced, and the system performance is ensured.

In one possible example, the brillouin optical signal reading and writing unit includes a writing optical waveguide, a reading optical waveguide, a photoacoustic conversion waveguide, a phonon free propagation region, and an acousto-optic conversion waveguide; the method further comprises the steps of:

In one possible example, a second read optical signal is also received by the read optical waveguide and input into the acousto-optic conversion waveguide; and realizing a nonlinear mapping function by combining the acousto-optic conversion waveguide with the second read optical signal and the phonon signal, and obtaining a first output optical signal.

In one possible example, the apparatus further comprises a first laser, a second laser, a modulator, an optical beam splitter, the first laser being connected to the modulator, the modulator being connected to the optical beam splitter, the second laser being connected to the brillouin optical signal reading and writing unit; the method further comprises the steps of:

Generating a first beam by a first laser;

loading a first signal to be processed on the first light beam through a modulator to obtain a first optical signal;

dividing the first optical signal into two paths of optical signals through an optical beam splitter, wherein one path of optical signal is a writing optical signal, and the other path of optical signal is a first reading optical signal;

a second beam is generated by a second laser, and the second beam is used as a second reading optical signal.

In one possible example, the apparatus further comprises a first optical combiner; the first optical combiner is connected with the Brillouin optical signal reading and writing unit; the method further comprises the steps of:

and receiving the first output optical signal and the second output optical signal through the first optical combiner, and linearly combining the first output optical signal and the second output optical signal to obtain a combined optical signal.

In one possible example, the apparatus further includes a third laser and a second optical combiner connected to the third laser, the second optical combiner further connected to the optical beam splitter and the brillouin optical signal reading and writing unit; the method further comprises the steps of:

generating a third light beam as a control light signal by a third laser;

and receiving a control optical signal and a writing optical signal through a second optical combiner, and inputting the control optical signal and the writing optical signal into a Brillouin optical signal reading and writing unit, wherein the control optical signal is used for providing energy for converting the writing optical signal into a phonon signal.

In one possible example, the first optical combiner is also connected to a detector, the detector being connected to the computing device; the method further comprises the steps of:

detecting the combined optical signal by a detector and converting the combined optical signal into a combined electrical signal;

and acquiring the combined electric signal through a computing device, acquiring an NGRC characteristic space according to the combined electric signal, and multiplying the NGRC characteristic space by an NGRC weight matrix to acquire an NGRC prediction result.

Therefore, in the embodiment of the application, a complete optical signal processing device is obtained through the combined connection of a series of components, the device can completely perform the optical method implementation process of the NGRC algorithm, and the device has the advantages of simple structure, low requirement on the device manufacturing process and high processing efficiency, and ensures the system performance realized by the optical method of the NGRC algorithm. Furthermore, the device can realize customization of NGRC algorithm by adjusting the length of the phonon free propagation region, the combination of the writing optical signal and the reading optical signal, thereby being applicable to different application scenes and practical problems.

As shown in fig. 10, fig. 10 shows a schematic hardware configuration of a communication device in an embodiment of the present application. Fig. 6A to 6D, fig. 7A to 7B, and fig. 8 correspond to the communication apparatus in the embodiment, or the structure of the communication apparatus for performing the method of the corresponding embodiment of fig. 9 may refer to the structure shown in fig. 10. The communication apparatus 800 includes: a processor 111 and a transceiver 112, the processor 111 and the transceiver 112 being electrically coupled;

The processor 111 is configured to execute some or all of the computer program instructions in the memory, which when executed, cause the apparatus to perform the method according to any of the embodiments described above.

A transceiver 112 for communicating with other devices; the writing optical signal, the first reading optical signal and the second reading optical signal are received by a brillouin optical signal reading and writing unit, for example, wherein the writing optical signal and the second reading optical signal bear a first signal to be processed.

Optionally, a memory 113 is further included for storing computer program instructions, optionally the memory 113 (memory # 1) is located within the device, the memory 113 (memory # 2) is integrated with the processor 111, or the memory 113 (memory # 3) is located outside the device.

It should be appreciated that the communication device 800 shown in fig. 10 may be a chip or a circuit. Such as a chip or circuit, which may be provided within the terminal device or communication device. The transceiver 112 may also be a communication interface. The transceiver includes a receiver and a transmitter. Further, the communication device 800 may also include a bus system.

The processor 111, the memory 113, and the transceiver 112 are connected through a bus system, where the processor 111 is configured to execute instructions stored in the memory 113 to control the transceiver to receive signals and send signals, so as to complete steps of the first device or the second device in the implementation method related to the present application. The memory 113 may be integrated in the processor 111 or may be provided separately from the processor 111.

As an implementation, the functions of the transceiver 112 may be considered to be implemented by a transceiver circuit or a transceiver-specific chip. The processor 111 may be considered to be implemented by a dedicated processing chip, a processing circuit, a processor or a general-purpose chip. The processor may be a central processor (central processing unit, CPU), a network processor (network processor, NP) or a combination of CPU and NP. The processor may further comprise a hardware chip or other general purpose processor. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (programmable logic device, PLD), or a combination thereof. The PLD may be a complex programmable logic device (complex programmable logic device, CPLD), a field-programmable gate array field-programmable gate array (FPGA), general-purpose array logic (generic array logic, GAL), and other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like, or any combination thereof. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

It should also be understood that the memory referred to in the embodiments of the present application may be volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random Access Memory, RAM) which acts as an external cache. By way of example, and not limitation, many forms of RAM are available, such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (Double Data Rate SDRAM), enhanced SDRAM (ESDRAM), synchronous DRAM (SLDRAM), and Direct RAM (DR RAM). It should be noted that the memory described herein is intended to comprise, without being limited to, these and any other suitable types of memory.

The present application provides a computer storage medium storing a computer program comprising instructions for performing the corresponding method of the above embodiments.

Embodiments of the present application provide a computer program product comprising instructions which, when run on a computer, cause the computer to perform the corresponding method of the above embodiments.

It should be understood that, in various embodiments of the present application, the sequence numbers of the foregoing processes do not mean the order of execution, and the order of execution of the processes should be determined by the functions and internal logic thereof, and should not constitute any limitation on the implementation process of the embodiments of the present application.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, and are not repeated herein.

In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.

The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The foregoing is merely specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily think about changes or substitutions within the technical scope of the present application, and the changes and substitutions are intended to be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An optical signal processing apparatus, characterized in that the apparatus comprises a brillouin optical signal reading and writing unit:

the brillouin optical signal reading and writing unit is configured to receive a writing optical signal, a first reading optical signal, and a second reading optical signal, where the writing optical signal and the second reading optical signal carry a first signal to be processed;

the brillouin optical signal reading and writing unit is further configured to implement a linear dimension increasing function according to the writing optical signal and the first reading optical signal, and obtain a first output optical signal;

2. The apparatus according to claim 1, wherein the brillouin optical signal reading and writing unit includes a writing optical waveguide, a reading optical waveguide, a photoacoustic conversion waveguide, a phonon free propagation region, and an acousto-optic conversion waveguide;

the writing optical waveguide is used for receiving the writing optical signal and inputting the writing optical signal into the photoacoustic conversion waveguide;

the photoacoustic conversion waveguide is used for converting the writing optical signal into a phonon signal;

The phonon free propagation region is used for propagating the phonon signal from the photoacoustic conversion waveguide to the acousto-optic conversion waveguide;

the reading optical waveguide is used for receiving the first reading optical signal and inputting the first reading optical signal into the acousto-optic conversion waveguide;

the acousto-optic conversion waveguide is used for combining the first read optical signal and the phonon signal to realize a linear dimension-increasing function, so as to obtain the first output optical signal.

3. The apparatus of claim 2, wherein the device comprises a plurality of sensors,

the reading optical waveguide is further configured to receive the second reading optical signal, and input the second reading optical signal into the acousto-optic conversion waveguide;

the acousto-optic conversion waveguide is further configured to combine the second read optical signal and the phonon signal to implement a nonlinear mapping function, so as to obtain the first output optical signal.

4. A device according to any one of claims 1-3, characterized in that the device further comprises a first laser, a second laser, a modulator, an optical beam splitter, the first laser being connected to the modulator, the modulator being connected to the optical beam splitter, the second laser being connected to the brillouin optical signal reading and writing unit;

The first laser is used for generating a first light beam;

the modulator is used for loading the first signal to be processed on the first light beam to obtain a first optical signal;

the optical beam splitter is configured to split the first optical signal into two optical signals, where one optical signal is the writing optical signal and the other optical signal is the first reading optical signal;

the second laser is used for generating a second light beam, and the second light beam is used as the second reading optical signal.

5. The apparatus of any one of claims 1-4, further comprising a first light combiner; the first optical combiner is connected with the Brillouin optical signal reading and writing unit;

the first optical combiner is configured to receive the first output optical signal and the second output optical signal, and perform linear combination on the first output optical signal and the second output optical signal to obtain a combined optical signal.

6. The apparatus of claim 5, further comprising a third laser and a second optical combiner coupled to the third laser, the second optical combiner further coupled to the optical splitter and the brillouin optical signal pickup unit;

The third laser is used for generating a third light beam as a control light signal;

the second optical combiner is configured to receive the control optical signal and the writing optical signal, and input the control optical signal and the writing optical signal into the brillouin optical signal reading and writing unit, where the control optical signal is used to provide energy for converting the writing optical signal into the phonon signal.

7. The apparatus of claim 5 or 6, wherein the first optical combiner is further coupled to a detector, the detector coupled to a computing device;

the computing device is used for obtaining the combined electrical signal, obtaining an NGRC characteristic space according to the combined electrical signal, multiplying the NGRC characteristic space by an NGRC weight matrix, and obtaining an NGRC prediction result.

8. An optical signal processing system comprising N parallel optical signal processing devices according to any of claims 1-7.

9. The system of claim 8, wherein the system comprises a brillouin optical signal read-write array, the brillouin optical signal read-write array being composed of N brillouin optical signal read-write units.

10. The system of claim 9, further comprising N first lasers and N modulators respectively connected to the N first lasers, the N first lasers being configured to generate N first light beams, the N modulators being configured to load first signals to be processed for the N first light beams respectively, to obtain N first optical signals;

the system further includes a second laser for generating a second beam of light, the second beam of light being a second read optical signal;

the N first optical signals and the second read optical signals are combined into one optical signal through a first optical combiner, the one optical signal is divided into two optical signals through a first optical beam splitter, a first optical signal in the two optical signals passes through an array waveguide grating, the array waveguide grating is used for dividing the first optical signal into N write optical signals according to data signals, and a second optical signal in the two optical signals is divided into N first read optical signals corresponding to the data signals through a second optical beam splitter;

and respectively inputting the N paths of writing optical signals, the N paths of first reading optical signals and the N paths of second reading optical signals into the N Brillouin optical signal reading and writing units, wherein one path of writing optical signals and one path of data signals of the first reading optical signals which are input by each Brillouin optical signal reading and writing unit correspond to each other.

11. The system of claim 10, further comprising a third laser, and a third optical splitter coupled to the third laser, the third optical splitter further coupled to N second optical combiners, each of the N second optical combiners coupled to one of the brillouin optical signal writing and reading units,

the third laser is used for generating a third light beam as a control light signal, the third light beam splitter receives the control light signal and divides the control light signal into N paths of control light signals, and the N paths of control light signals are respectively input into the N second light beam combiners;

each of the N second optical combiners receives one of N writing optical signals, and connects the writing optical signal with the brillouin optical signal reading and writing unit connected to the control optical signal input, where the control optical signal is used to provide energy for converting the writing optical signal into the phonon signal.

12. The system of claim 11, wherein when the second optical splitter splits the control optical signal into N control optical signals, the control optical signal splitting ratio is determined according to weights of the N brillouin optical signal reading and writing units.

13. An optical signal processing method, wherein the method is applied to an optical signal processing device, the device comprises a brillouin optical signal reading and writing unit, and the method comprises:

receiving a writing optical signal, a first reading optical signal and a second reading optical signal through the Brillouin optical signal reading and writing unit, wherein the writing optical signal and the second reading optical signal bear a first signal to be processed;

14. The method of claim 13, wherein the brillouin optical signal read-write unit comprises a write optical waveguide, a read optical waveguide, a photoacoustic conversion waveguide, a phonon free propagation region, and an acousto-optic conversion waveguide; the method further comprises the steps of:

receiving the writing optical signal through the writing optical waveguide, and inputting the writing optical signal into the photoacoustic conversion waveguide;

Converting the writing optical signal into a phonon signal through the photoacoustic conversion waveguide;

propagating the phonon signal from the photoacoustic conversion waveguide to the acousto-optic conversion waveguide through the phonon free propagation region;

receiving the first reading optical signal through the reading optical waveguide, and inputting the first reading optical signal into the acousto-optic conversion waveguide;

and the acousto-optic conversion waveguide is combined with the first read optical signal and the phonon signal to realize a linear dimension-lifting function, so that the first output optical signal is obtained.

15. The method of claim 14, wherein the method further comprises:

receiving the second reading optical signal through the reading optical waveguide, and inputting the second reading optical signal into the acousto-optic conversion waveguide;

and combining the second read optical signal and the phonon signal through the acousto-optic conversion waveguide to realize a nonlinear mapping function, so as to obtain the first output optical signal.

16. The method according to any one of claims 13-15, wherein the apparatus further comprises a first laser, a second laser, a modulator, an optical beam splitter, the first laser being connected to the modulator, the modulator being connected to the optical beam splitter, the second laser being connected to the brillouin optical signal reading and writing unit; the method further comprises the steps of:

Generating a first beam by the first laser;

loading the first signal to be processed on the first light beam through the modulator to obtain a first optical signal;

dividing the first optical signal into two paths of optical signals through the optical beam splitter, wherein one path of optical signal is the writing optical signal, and the other path of optical signal is the first reading optical signal;

generating a second light beam by the second laser, and taking the second light beam as the second reading light signal.

17. The method of any one of claims 13-16, wherein the apparatus further comprises a first light combiner; the first optical combiner is connected with the Brillouin optical signal reading and writing unit; the method further comprises the steps of:

18. The method of claim 17, wherein the apparatus further comprises a third laser and a second optical combiner coupled to the third laser, the second optical combiner further coupled to the optical splitter and the brillouin optical signal pickup unit; the method further comprises the steps of:

Generating a third light beam as a control light signal by the third laser;

and receiving the control optical signal and the writing optical signal through the second optical combiner, and inputting the control optical signal and the writing optical signal into the brillouin optical signal reading and writing unit, wherein the control optical signal is used for providing energy for converting the writing optical signal into the phonon signal.

19. The method of claim 17 or 18, wherein the first optical combiner is further coupled to a detector, the detector coupled to a computing device; the method further comprises the steps of:

detecting the combined optical signal by the detector and converting the combined optical signal into a combined electrical signal;

and acquiring the combined electric signal through the computing equipment, acquiring an NGRC characteristic space according to the combined electric signal, and multiplying the NGRC characteristic space by an NGRC weight matrix to acquire an NGRC prediction result.

20. A method of optical signal processing, characterized in that the method comprises steps for performing the functions of the system as claimed in any of claims 8-12.

21. A communication device, wherein the device comprises a processor and may further comprise a memory; the processor is coupled to the memory and operable to execute computer program instructions stored in the memory to cause an apparatus to perform the method of any one of claims 13-19 or to cause an apparatus to perform the method of claim 20.

22. A readable storage medium storing instructions which, when executed, cause the method of any one of claims 13-19 to be implemented or cause the method of claim 20 to be implemented.