CN116107037A - Optical computing network structure of optical device on chip of dense waveguide array - Google Patents
Optical computing network structure of optical device on chip of dense waveguide array Download PDFInfo
- Publication number
- CN116107037A CN116107037A CN202310389774.7A CN202310389774A CN116107037A CN 116107037 A CN116107037 A CN 116107037A CN 202310389774 A CN202310389774 A CN 202310389774A CN 116107037 A CN116107037 A CN 116107037A
- Authority
- CN
- China
- Prior art keywords
- optical
- waveguide array
- dense waveguide
- dense
- network structure
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G02—OPTICS
- G02B—OPTICAL ELEMENTS, SYSTEMS OR APPARATUS
- G02B6/00—Light guides; Structural details of arrangements comprising light guides and other optical elements, e.g. couplings
- G02B6/24—Coupling light guides
- G02B6/26—Optical coupling means
- G02B6/35—Optical coupling means having switching means
- G02B6/354—Switching arrangements, i.e. number of input/output ports and interconnection types
-
- G—PHYSICS
- G02—OPTICS
- G02B—OPTICAL ELEMENTS, SYSTEMS OR APPARATUS
- G02B6/00—Light guides; Structural details of arrangements comprising light guides and other optical elements, e.g. couplings
- G02B6/10—Light guides; Structural details of arrangements comprising light guides and other optical elements, e.g. couplings of the optical waveguide type
- G02B6/12—Light guides; Structural details of arrangements comprising light guides and other optical elements, e.g. couplings of the optical waveguide type of the integrated circuit kind
- G02B6/12007—Light guides; Structural details of arrangements comprising light guides and other optical elements, e.g. couplings of the optical waveguide type of the integrated circuit kind forming wavelength selective elements, e.g. multiplexer, demultiplexer
-
- G—PHYSICS
- G02—OPTICS
- G02B—OPTICAL ELEMENTS, SYSTEMS OR APPARATUS
- G02B6/00—Light guides; Structural details of arrangements comprising light guides and other optical elements, e.g. couplings
- G02B6/24—Coupling light guides
- G02B6/26—Optical coupling means
- G02B6/28—Optical coupling means having data bus means, i.e. plural waveguides interconnected and providing an inherently bidirectional system by mixing and splitting signals
- G02B6/293—Optical coupling means having data bus means, i.e. plural waveguides interconnected and providing an inherently bidirectional system by mixing and splitting signals with wavelength selective means
- G02B6/29346—Optical coupling means having data bus means, i.e. plural waveguides interconnected and providing an inherently bidirectional system by mixing and splitting signals with wavelength selective means operating by wave or beam interference
-
- 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
- H04B10/50—Transmitters
- H04B10/516—Details of coding or modulation
- H04B10/54—Intensity modulation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Optics & Photonics (AREA)
- Engineering & Computer Science (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Electromagnetism (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Optical Modulation, Optical Deflection, Nonlinear Optics, Optical Demodulation, Optical Logic Elements (AREA)
Abstract
The invention relates to the technical field of photoelectric chips, in particular to an optical computing network structure of an on-chip optical device of a dense waveguide array. Comprising the following steps: a modulatable dense waveguide array, a power intensity modulator; the two groups of the modulatable dense waveguide arrays are connected through a power intensity modulator; the multipath optical signal vector is input from one side of the device, transmitted to the first modulating dense waveguide array and modulated to redistribute the splitting ratio; modulating the power intensity of each path of signal by a power modulator; after modulation, inputting a second modulating dense waveguide array, and outputting the optical signal vector after further optical signal distribution; the optical signal vector output through the whole network structure and the input multipath optical signal vector realize any linear transformation relation; the network structure is a U x S x V matrix. The device has the advantages of small size, effectively reduced size of the large-scale transmission matrix, contribution to large-scale integration, higher modulation speed and higher matrix change realization speed.
Description
Technical Field
The invention relates to the technical field of photoelectric chips, in particular to an optical computing network structure of an on-chip optical device of a dense waveguide array.
Background
Linear matrix multiplication or multiply-accumulate is a subject computing task for parallel computing systems, especially for image processing or deep learning techniques, where more than 90% of the computing tasks belong to this class of computation for deep learning algorithms. Therefore, the method improves the calculation speed and calculation capacity of the hardware system for linear matrix multiplication and reduces the calculation energy consumption, and is a core research problem of modern neuron network calculation.
With the trend of moore's law towards the end of technology, the traditional electric signal transmission has unavoidable problems of delay limit, bandwidth limit and high energy consumption, and the computing performance of the pure electronic chip is approaching to the performance bottleneck in the near future. The optical network is utilized to realize linear matrix multiplication so as to accelerate the whole system, thereby becoming a feasible scheme for breaking through the technical bottleneck of the existing deep network hardware.
The current technical scheme for realizing linear matrix multiplication by an optical mode mainly comprises the following steps: (1) Multi-plane optical signal conversion (Multi-Plane Light Conversion, MPLC) (e.g., D2NN structure); (2) The broadcast signal is then weighted (Broadcast and Weight) (e.g., a wavelength division multiplexing network); (3) Power splitting (e.g., MZI networks). A simple structure of several technical solutions is schematically shown in fig. 1. All the above schemes have the problems that the whole system is large in size, and a large-scale and high-density computing optical network is difficult to integrate on a single chip. Wherein the MPLC-based solution requires the introduction of multiple parallel diffraction planes, and is completely incapable of monolithic integration. While both types of schemes based on MZI grids and wavelength division multiplexing can be integrated on a single chip, large scale integration is difficult due to the large size of the on-chip optical elements that make up the system, and the large number of such element devices required for large scale linear matrix multiplication networks.
Disclosure of Invention
The present invention is directed to solving the above-mentioned problems, and provides an optical computing network structure of a multiple-input-multiple-output on-chip optical device structure based on a tunable dense waveguide array with dynamic spectral ratio.
The invention aims to provide an optical computing network structure of an on-chip optical device of a dense waveguide array, which comprises the following components: two groups of modulatable dense waveguide arrays, a group of power intensity modulators; the two groups of the modulatable dense waveguide arrays are connected through a power intensity modulator;
the multipath optical signal vector is input from one side, transmitted to a first modulating dense waveguide array and modulated to redistribute the splitting ratio; modulating the power intensity of each path of signal through a power modulator; after modulation, inputting a second modulating dense waveguide array, and outputting the optical signal vector after further optical signal distribution; the optical signal vector output through the whole network structure and the input multipath optical signal vector realize any linear transformation relation.
Preferably, the on-chip optics optical computing network structure of the dense waveguide array is a U x S x V matrix, where U and V are unitary matrices and S is a diagonal matrix.
Preferably, the network structure further comprises an input port array and an output port array, and the input port array, the modulatable dense waveguide array, the power intensity modulator, the modulatable dense waveguide array and the output port array are sequentially connected in series; the multipath optical signal vectors are input by the input port array, and the output optical signal vectors are output by the output port array.
Preferably, the power intensity modulator is a light intensity modulator for modulating the intensity of the optical signal.
Preferably, the light intensity modulator includes an optical switching device, an optical absorption modulator, and an optical amplifier.
Preferably, the optical switching device is a MZI optical switching device.
Preferably, the optical absorption modulator is an electroabsorption optical modulator.
Preferably, the optical amplifier is a semiconductor optical amplifier.
The invention has the beneficial effects that:
compared with the prior MZI grid linear matrix multiplication accelerator scheme based on power beam splitting, the invention integrates a large number of discrete large-size optical units in a power beam splitting network into a single complex on-chip optical device, the modulatable dense waveguide array can effectively reduce the size of a large-scale transmission matrix, and the trend is further remarkable along with the scale growth of the linear matrix; when the scale of the linear matrix reaches 1024×1024, the device volume required by the invention is only one percent of the device volume in the prior MZI grid technical scheme, which plays a significant role in promoting the large-scale integration of the optical matrix multiplier. Compared with the existing thermo-optical modulation mode, the method has the advantages that the modulation speed is higher, and the matrix change is realized more quickly.
Drawings
Fig. 1 is a schematic diagram of a conventional optical network structure; (a) MPLC-based optical network architecture; (b) an optical network architecture based on wavelength division multiplexing; (c) an optical network structure based on an MZI mesh.
FIG. 2 is a schematic diagram of an optical computing network structure of an on-chip optical device of a dense waveguide array according to an embodiment of the present invention; v and U represent unitary matrices and S represents diagonal matrices.
Reference numerals:
1. an input port array; 2. a modulating dense waveguide array; 3. a power intensity modulator; 4. an array of output ports.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not to be construed as limiting the invention.
The invention provides an optical computing network structure of an on-chip optical device of a dense waveguide array, which comprises the following components: two groups of modulatable dense waveguide arrays, a group of power intensity modulators; the two groups of the modulatable dense waveguide arrays are connected through a power intensity modulator;
the multipath optical signal vector is input from one side, transmitted to a first modulating dense waveguide array and modulated to redistribute the splitting ratio; modulating the power intensity of each path of signal through a power modulator; after modulation, inputting a second modulating dense waveguide array, and outputting the optical signal vector after further optical signal distribution; the optical signal vector output through the whole network structure and the input multipath optical signal vector realize any linear transformation relation;
the optical computing network structure of the on-chip optical devices of the dense waveguide array is a U×S×V matrix, wherein U and V are unitary matrices, and S is a diagonal matrix;
the network structure also comprises an input port array and an output port array, wherein the input port array, the modulatable dense waveguide array, the power intensity modulator, the modulatable dense waveguide array and the output port array are sequentially connected in series; the multipath optical signal vectors are input by the input port array, and the output optical signal vectors are output by the output port array;
the power intensity modulator is a light intensity modulator for modulating the intensity of the optical signal.
The light intensity modulator comprises an optical switching device, an optical absorption modulator and an optical amplifier;
preferably, the optical switching device is an MZI optical switching device;
preferably, the optical absorption modulator is an electroabsorption optical modulator;
preferably, the optical amplifier is a semiconductor optical amplifier.
The front and back two modulatable dense waveguide arrays in the network only change the output signal splitting ratio by modulating the phase of the optical signal, the intensity of the optical signal is not changed, and a group of light intensity modulators connected with the two dense waveguide array devices in series only modulate the intensity of the optical signal, and the optical phase is not obviously influenced; the light intensity modulator may be implemented by a combination of various optical switching devices, optical absorption modulators, and optical amplifiers; on-chip optical switches are typically implemented by MZI optical switching devices, optical absorption modulators are typically implemented by Electro-absorption modulator (EAM) effect, and optical amplifiers are typically implemented by hybrid integrated on-chip semiconductor optical amplifiers.
Example 1
For a typical lossless optical network, such as a MZI network, the multiple input-output optical field intensity transmission matrix is typically a unitary matrix, i.e., the conjugate transpose of the matrix is its inverse. To achieve random adjustment of the multiple input-output power ratios, singular value decomposition (Single Valued Decomposition, SVD) is required to decompose any linear matrix into a matrix multiplication of uxxsxv, where the U and V matrices are unitary and S is a diagonal matrix. It can be proved that the optical field intensity transmission matrix based on the multi-input-multi-output on-chip optical device structure of the optical density waveguide array with the dynamically adjustable splitting ratio is also a unitary matrix, and the transmission matrix of the device can be modulated into any unitary matrix by modulating voltages or currents at different positions on the device, which is similar to the pure MZI network function. Therefore, the functions of a U matrix and a V matrix can be realized based on a multi-input-multi-output on-chip optical device structure of the dynamic adjustable dense waveguide array of the splitting ratio, and the S matrix is realized by connecting an optical switch, an electroabsorption optical modulator and a semiconductor optical amplifier in series; the overall network structure is shown in fig. 2.
An on-chip optical device optical computing network structure of a dense waveguide array is shown, and comprises an input port array 1, two groups of modulatable dense waveguide arrays 2, a group of power intensity modulators 3 and an output port array 4;
the input port array 1, the modulatable dense waveguide array 2, the power intensity modulator 3, the modulatable dense waveguide array 2 and the output port array 4 are sequentially connected in series;
the power intensity modulator 3 is a light intensity modulator; the light intensity modulator comprises an MZI optical switching device, an electroabsorption light modulator and a semiconductor optical amplifier;
the multi-path optical signal vector is input by an input port array 1 of a network structure, is transmitted to a first modulatable dense waveguide array 2, is subjected to modulation and redistribution of a splitting ratio by a device, is subjected to modulation by a group of power modulators 3, is subjected to power intensity modulation of each path of signal, and is continuously input to a second modulatable dense waveguide array 2, and further optical signal distribution is carried out on the optical signal vector; output from the output port array 4; finally, a group of optical signal vectors passing through the whole optical network and the input optical signal vector can realize any linear transformation relation;
the front and back two modulatable dense waveguide arrays 2 in the network only change the output signal splitting ratio by modulating the phase of the optical signal, and do not change the intensity of the optical signal, while a group of light intensity modulators connected with the two dense waveguide arrays 2 in series only modulate the intensity of the optical signal, and do not have significant influence on the optical phase.
While embodiments of the present invention have been illustrated and described above, it will be appreciated that the above described embodiments are illustrative and should not be construed as limiting the invention. Variations, modifications, alternatives and variations of the above-described embodiments may be made by those of ordinary skill in the art within the scope of the present invention.
The above embodiments of the present invention do not limit the scope of the present invention. Any other corresponding changes and modifications made in accordance with the technical idea of the present invention shall be included in the scope of the claims of the present invention.
Claims (8)
1. An on-chip optical device optical computing network architecture of a dense waveguide array, comprising: two groups of modulatable dense waveguide arrays, a group of power intensity modulators; the two groups of the modulatable dense waveguide arrays are connected through a power intensity modulator;
the multipath optical signal vector is input from one side, transmitted to a first modulating dense waveguide array and modulated to redistribute the splitting ratio; modulating the power intensity of each path of signal through a power modulator; after modulation, inputting a second modulating dense waveguide array, and outputting the optical signal vector after further optical signal distribution; the optical signal vector output through the whole network structure and the input multipath optical signal vector realize any linear transformation relation.
2. The dense waveguide array on-chip optics optical computing network architecture of claim 1 wherein: the optical computing network structure of the on-chip optical devices of the dense waveguide array is a U×S×V matrix, wherein U and V are unitary matrices, and S is a diagonal matrix.
3. The dense waveguide array on-chip optics optical computing network architecture of claim 2 wherein: the network structure further comprises an input port array and an output port array; the input port array, the modulatable dense waveguide array, the power intensity modulator, the modulatable dense waveguide array and the output port array are sequentially connected in series; the multipath optical signal vectors are input by the input port array, and the output optical signal vectors are output by the output port array.
4. An on-chip optics optical computing network structure of a dense waveguide array according to claim 3, characterized in that: the power intensity modulator is a light intensity modulator for modulating the intensity of the optical signal.
5. The dense waveguide array on-chip optics optical computing network structure of claim 4 wherein: the light intensity modulator includes an optical switching device, an optical absorption modulator, and an optical amplifier.
6. The dense waveguide array on-chip optics optical computing network structure of claim 5 wherein: the optical switching device is an MZI optical switching device.
7. The dense waveguide array on-chip optical computing network structure of claim 5 or 6, wherein: the optical absorption modulator is an electroabsorption optical modulator.
8. The dense waveguide array on-chip optics optical computing network structure of claim 7 wherein: the optical amplifier is a semiconductor optical amplifier.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310389774.7A CN116107037B (en) | 2023-04-13 | 2023-04-13 | Optical computing network structure of optical device on chip of dense waveguide array |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310389774.7A CN116107037B (en) | 2023-04-13 | 2023-04-13 | Optical computing network structure of optical device on chip of dense waveguide array |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116107037A true CN116107037A (en) | 2023-05-12 |
| CN116107037B CN116107037B (en) | 2023-07-07 |
Family
ID=86264152
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310389774.7A Active CN116107037B (en) | 2023-04-13 | 2023-04-13 | Optical computing network structure of optical device on chip of dense waveguide array |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116107037B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20010010739A1 (en) * | 2000-01-06 | 2001-08-02 | Koichi Takiguchi | CDMA encoder-decoder, CDMA communication system, WDM-CDMA communication system |
| CN111683304A (en) * | 2020-05-13 | 2020-09-18 | 中国科学院西安光学精密机械研究所 | All-optical diffraction neural network and system realized on optical waveguide and/or optical chip |
| US20210064958A1 (en) * | 2020-11-17 | 2021-03-04 | Intel Corporation | Heterogeneously integrated optical neural network accelerator |
| CN112986957A (en) * | 2021-03-16 | 2021-06-18 | 长沙思木锐信息技术有限公司 | Beam guide wiring structure, optical scanning device and calibration method thereof |
| CN113961035A (en) * | 2021-10-15 | 2022-01-21 | 北京理工大学 | Deep learning-oriented high-efficiency reconfigurable all-optical neural network computing chip architecture |
| WO2022262147A1 (en) * | 2021-06-17 | 2022-12-22 | 苏州浪潮智能科技有限公司 | Optical device, electronic device, and programmable photonic integrated circuit |
-
2023
- 2023-04-13 CN CN202310389774.7A patent/CN116107037B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20010010739A1 (en) * | 2000-01-06 | 2001-08-02 | Koichi Takiguchi | CDMA encoder-decoder, CDMA communication system, WDM-CDMA communication system |
| CN111683304A (en) * | 2020-05-13 | 2020-09-18 | 中国科学院西安光学精密机械研究所 | All-optical diffraction neural network and system realized on optical waveguide and/or optical chip |
| US20210064958A1 (en) * | 2020-11-17 | 2021-03-04 | Intel Corporation | Heterogeneously integrated optical neural network accelerator |
| CN112986957A (en) * | 2021-03-16 | 2021-06-18 | 长沙思木锐信息技术有限公司 | Beam guide wiring structure, optical scanning device and calibration method thereof |
| WO2022262147A1 (en) * | 2021-06-17 | 2022-12-22 | 苏州浪潮智能科技有限公司 | Optical device, electronic device, and programmable photonic integrated circuit |
| CN113961035A (en) * | 2021-10-15 | 2022-01-21 | 北京理工大学 | Deep learning-oriented high-efficiency reconfigurable all-optical neural network computing chip architecture |
Non-Patent Citations (2)
| Title |
|---|
| GEORGE GIAMOUGIANNIS.ET.AL: "A Coherent Photonic Crossbar for Scalable Universal Linear Optics", JOURNAL OF LIGHTWAVE TECHNOLOGY, vol. 41, no. 8 * |
| SHAOFU XU.ET.AL: "Parallel optical coherent dot-product architecture for large-scale matrix multiplication with compatibility for diverse phase shifters", OPTICS EXPRESS, vol. 30, no. 23 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116107037B (en) | 2023-07-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9485048B2 (en) | Methods and devices for space-time multi-plane optical networks | |
| US6700517B1 (en) | Photonic analog-to-digital converter | |
| CN112099048B (en) | Microwave photon MIMO radar detection method and system based on time division-difference frequency multiplexing | |
| CN114815959B (en) | Photon tensor calculation acceleration method and device based on wavelength division multiplexing | |
| CN114358271A (en) | Time-wavelength interweaving photon neural network convolution acceleration chip | |
| CN115222035B (en) | Photon neural network convolution acceleration chip | |
| WO2022164746A1 (en) | Photonic accelerator for deep neural networks | |
| CN116432726B (en) | Photoelectric hybrid deep neural network operation device and operation method | |
| Tucker | Optical packet switching: A reality check | |
| CN116029350B (en) | Two-dimensional photon coherent convolution acceleration chip based on time interleaving and application system thereof | |
| Tsakyridis et al. | Universal linear optics for ultra-fast neuromorphic silicon photonics towards Fj/MAC and TMAC/sec/mm 2 engines | |
| CN117436486A (en) | Optical convolution neural network based on thin film lithium niobate and silicon mixture | |
| CN116107037B (en) | Optical computing network structure of optical device on chip of dense waveguide array | |
| CN116484931B (en) | Photon matrix multiplication operation device and operation method for neural network | |
| KR20230042333A (en) | Systems and methods for exploiting photonic degrees of freedom in a photonic processor | |
| US11934943B1 (en) | Two-dimensional photonic neural network convolutional acceleration chip based on series connection structure | |
| Calabretta et al. | Monolithically integrated WDM cross-connect switch for high-performance optical data center networks | |
| Wu et al. | SiP architecture for accelerating collective communication in distributed deep learning | |
| US20240022333A1 (en) | Photonics-aided vector terahertz signal communication system | |
| WO2021166173A1 (en) | Optical iq modulator | |
| Sorokina | High bandwidth all-optical fiber-based neuromorphic signal processing | |
| Fayyaz et al. | Performance analysis of optical interconnects’ architectures for data center networks: Do you have a subtitle? If so, write it here | |
| Yuanzhi et al. | Satellite Communication Payload Based on Microwave Photonics: Benefits, Architecture, and Technologies | |
| CN117709423B (en) | Deep neural network photon acceleration chip and operation system thereof | |
| CN116306858A (en) | Neural network device based on multimode interference and operation method thereof |
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 |