Photonic 2D Convolution Using Wavelength-Time Interleaving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current electronic neural networks face challenges with high power consumption, long delays, and limited speed due to their classical computer structure, which separates program and data spaces, leading to inefficient data workflow and computational power usage.
Innovation Solution
A two-dimensional photonic convolutional acceleration system utilizing a dispersion module, optical fiber delay array, and microring weighting array chip to perform two-level delay and signal strength weighting, enabling two-dimensional convolutional kernel matrix coefficient weighting in a single signal cycle, addressing data redundancy and multi-dimensional data convolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If traditional electronic chip-based neural networks are used, then computational operations can be performed, but power consumption is high and computational speed is limited
Solution Approach 1:
The patent replaces electronic computation with photonic computation. The optical processing system uses light to perform convolution operations instead of electronic circuits, achieving lower power consumption while maintaining high computational speed. The optical domain processing eliminates the need for electron-based signal processing, directly addressing the contradiction between power consumption and computational speed.
Solution Approach 2:
The patent changes the fundamental parameter of information carrier from electrical signals to optical signals. By using light with its higher frequency and lower attenuation characteristics, the system achieves both reduced power consumption and improved computational speed simultaneously, resolving the trade-off present in electronic systems.
2Reliability
If data workflow between storage unit and computing unit is separated, then computational operations can be performed, but data workflow becomes unstable and power consumption increases
Solution Approach 1:
The patent merges the storage and computation functions into a single optical processing unit. The convolutional kernel parameters are stored within the optical processing system itself, eliminating the need for separate data transfer between storage and computing units. This integration stabilizes the data workflow and reduces the power consumption associated with data movement.
3Productivity
If traditional electronic chips are used, then computational operations can be performed, but micro-quantum characteristics and macro-high-frequency response characteristics create challenges
Solution Approach 1:
The patent substitutes photonic components for electronic chip components. By using optical waves instead of electrical signals, the system avoids the fundamental limitations of electronic chips related to micro-quantum effects and high-frequency response. This substitution simplifies the device architecture while improving computational efficiency.
4Productivity
If photonic technology is used for convolutional operations, then computational speed increases and power consumption decreases, but implementation complexity increases
Solution Approach 1:
The patent segments the photonic processing system into functional modules: optical signal generation, modulation, optical processing, and detection. Each module performs a specific function, making the overall complex system more manageable and implementable. This modular segmentation maintains the high computational speed benefits while reducing implementation complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system achieves efficient convolutional acceleration with reduced power consumption and increased bandwidth, suitable for real-time data processing by leveraging photonic technology's low loss and parallelizability, and supports flexible expansion of the convolutional kernel matrix.
Implementation Method 1
a multi-wavelength light source, generating multi-wavelength optical signals containing N wavelengths
Implementation Method 2
the dispersion module, realizing equal-interval dispersion delay for the N sub-modulated optical signals corresponding to N wavelengths
Implementation Method 3
the optical fiber delay array, which is composed of M-segment optical fibers, increasing the equal-interval second-level delay of M-channel multi-wavelength modulated optical signal
Implementation Method 4
the microring weighting array chip, comprising M microring weighting units, which are respectively used to weight and sum the N sub-modulated optical signals
Implementation Method 5
the modulator, loading the signal to be convolved onto the multi-wavelength optical signal, obtaining a multi-wavelength modulated optical signal
Implementation Method 6
the trans-impedance amplifier array, comprising M trans-impedance amplifiers, amplifying the M first-level weighted summation electrical signals
Data Source
AI summary
The present invention discloses a two-dimensional photonic convolutional acceleration system and device for convolutional neural network, comprising: a multi-wavelength light source, a signal source to be convolved, a modulator, a dispersion module, a 1×M power divider, an optical fiber delay array, a microring weighting array chip, a convolutional kernel matrix control unit, a trans-impedance amplifier array, and an acquisition and processing unit. The present invention realizes two-dimensional convolutional acceleration based on wavelength-time interleaving technology, a single modulator can realize the optical domain loading of the signal, and the convolutional operation speed is only limited to the speed of the modulator. The present invention can realize two-dimensional convolutional kernel convolutional acceleration of two-dimensional data in a single signal cycle based on two-level delay and microring weighting array chip, and solve the problem of data redundancy in traditional methods, the scheme is simple and efficient. The present invention realizes the control of convolutional kernel matrix coefficient based on the microring weighting array chip, can realize the fast update of convolutional kernel matrix coefficient, and is suitable for real-time data processing applications, the balanced photodetector can realize arbitrary convolutional kernel coefficient weighting.