Arithmetic Device Magnetic Flux Product-Sum Operations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hardware implementations of neural networks face challenges in simplifying arithmetic processing within individual units while performing product-sum operations efficiently.
Innovation Solution
An arithmetic device is designed with a positive-side current source, negative-side current source, cross switches, a coefficient memory unit, and a comparator, which performs product-sum operations and sign function processing to output a two-state value, simulating nonlinear operations with a simple configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a simple configuration is used for arithmetic processing in neural network units, then device complexity is reduced, but manufacturing precision and calculation accuracy may deteriorate
Solution Approach 1:
The patent replaces traditional voltage-based computation with a magnetic field-based system. Magnetic cores with hysteresis loops perform arithmetic operations through magnetic flux coupling, eliminating the need for complex voltage amplification and signal conditioning circuits. This substitution of physical domain (from electrical voltage to magnetic flux) enables simple, amplifier-free computation while maintaining calculation accuracy through the inherent stability of magnetic states.
Solution Approach 2:
The patent changes the fundamental operating parameter from voltage to magnetic flux density. By utilizing the B-H curve characteristics of magnetic materials, the system performs multiplication and addition operations through controlled magnetic coupling. The hysteresis loop of the magnetic core provides stable binary states (0 and 1) that are less susceptible to noise and drift, thereby maintaining precision while simplifying the overall device configuration.
2Measurement precision
If traditional voltage-based arithmetic operations are used, then calculation precision is maintained, but device complexity and energy consumption increase
Solution Approach 1:
The patent substitutes voltage-based arithmetic operations with magnetic flux-based operations. Magnetic cores coupled through mutual inductance perform multiplication and accumulation directly in the magnetic domain. This eliminates the need for voltage amplifiers, signal conditioners, and complex feedback circuits, thereby reducing device complexity while preserving arithmetic precision through the stability of magnetic hysteresis characteristics.
Solution Approach 2:
The patent extracts and removes unnecessary intermediate components from traditional voltage-based arithmetic circuits. By performing computations directly in the magnetic domain, the system eliminates amplifiers, comparators, and signal conditioning stages that would otherwise be required in voltage-based implementations. This extraction of redundant components reduces circuit complexity while maintaining calculation accuracy.
3Reliability
If voltage amplification circuits are added to improve signal strength, then signal quality is improved, but device complexity and energy consumption increase
Solution Approach 1:
The patent replaces voltage amplification circuits with direct magnetic flux coupling between cores. The magnetic signal is transmitted and processed through mutual inductance without requiring active amplification. The inherent gain in magnetic coupling, combined with the stability of hysteresis-based state retention, provides sufficient signal quality without adding amplifiers or active components, thereby reducing circuit complexity.
Solution Approach 2:
The magnetic core system performs self-amplification through the nonlinear hysteresis characteristics of the magnetic material. The B-H curve provides automatic signal regeneration and state stabilization without external amplification circuits. This self-service mechanism maintains signal quality and reliability while avoiding the need for additional active components that would increase device complexity and energy consumption.
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 device achieves efficient arithmetic processing by reducing the dynamic range of voltage differences input to the comparator, enabling a simple configuration for simulating neurons and performing multiply accumulate operations effectively.
Implementation Method 1
a positive-side current source configured to output a current from a positive-side terminal, and output a first voltage corresponding to a value of 1/L of the current output from the positive-side terminal
Implementation Method 2
a negative-side current source configured to output a current from a negative-side terminal, and output a second voltage corresponding to a value of I/L of the current output from the negative-side terminal
Implementation Method 3
a comparator configured to output an output signal having a value corresponding to a comparison result of the first voltage with the second voltage
Data Source
AI summary
According to an embodiment, an arithmetic device is configured to receive M input signals each representing a two-state value and M coefficients to output an output signal representing a two-state value. The device includes a positive-side current source, a negative-side current source, M cross switches, a coefficient memory unit, and a comparator. The positive-side current source is configured to output a first voltage corresponding to a value of 1/L of the current output from a positive-side terminal. The negative-side current source is configured to output a second voltage corresponding to a value of 1/L of the current output from a negative-side terminal. The memory unit includes M cells corresponding to the respective M coefficients. The comparator is configured to output an output signal having a value corresponding to a comparison result of the first voltage with the second voltage. Each M cell includes a first resistor and a second resistor.