Resistive Memory Artificial Neuron Circuit Design
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current artificial neuron implementations based on hybrid analog/digital VLSI circuits are not suitable for integration with emerging nanoscale devices like resistive memory cells, and there is a lack of a concrete proposal for realizing artificial neurons that capture the essential attributes of biological neurons for neuromorphic hardware.
Innovation Solution
An artificial neuron apparatus utilizing a resistive memory cell with a circuitry that includes input terminals for applying neuron signals, a read circuit for producing a read signal based on the cell's resistance, a storage circuit for storing measurement signals, and a switch set to reset the memory cell, allowing the neuron to accumulate input signals and 'fire' when reaching a specific resistance state, and then reset to its original state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If hybrid analog/digital VLSI circuits are used to realize artificial neurons, then the neurons can be integrated into existing CMOS technology, but the device complexity and number of transistors required increases significantly
Solution Approach 1:
The patent merges the neuron body, synapse, and memory functions into a single resistive memory cell, eliminating the need for separate transistor-based circuits. This consolidation reduces device complexity while maintaining manufacturability through standard resistive memory fabrication processes
Solution Approach 2:
The patent replaces the mechanical/electronic transistor-based VLSI circuit implementation with a physics-based resistive memory cell that uses electrical resistance changes to represent neuronal states. This substitution eliminates the need for complex transistor networks while enabling direct integration with emerging nanoscale devices
2Measurement precision
If conventional supercomputers are used to simulate brain activity, then computational accuracy can be maintained, but power consumption and space requirements increase dramatically
Solution Approach 1:
The resistive memory cell autonomously performs integration of input signals through its inherent capacitance and resistance characteristics, eliminating the need for external computational circuits. The cell naturally accumulates charge and reaches threshold states without requiring additional power-consuming processing units
Solution Approach 2:
The patent changes the operational parameters from high-power digital voltage switching to low-power analog resistance modulation. The resistive memory cell operates at much lower voltage levels and power consumption while maintaining computational functionality through resistance state changes rather than digital logic transitions
3Device complexity
If resistive memory cells are used to realize artificial neurons, then device complexity and chip area are reduced, but the ability to capture essential biological neuron attributes must be maintained
Solution Approach 1:
The resistive memory cell is designed to perform multiple neuronal functions simultaneously: it acts as the neuron body for signal integration, as a synapse for weighted connection, and as a memory element for state retention. This multi-functionality captures essential biological neuron attributes while minimizing device complexity
Solution Approach 2:
The patent implements dynamic behavior in the resistive memory cell through time-dependent resistance changes and threshold switching. The cell exhibits adaptive responses to input signals, including integration over time, threshold-based firing, and state transitions that mirror biological neuron dynamics without requiring complex circuitry
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
This solution enables the development of practical neuron realizations that can replace silicon neurons, reducing complexity and chip area, and are suitable for integration with other resistive memory cell-based components in neuromorphic systems, facilitating efficient neuromorphic network operation.
Implementation Method 1
These cells exhibit a threshold-switching effect whereby the cell can be switched between high and low resistance states by applying a control signal above a threshold level
Implementation Method 2
The cell-resistance can be measured (or 'read') by applying a low-voltage control signal to the electrodes and measuring the resulting current flow through the cell
Data Source
AI summary
A resistive memory cell is connected in circuitry which has a first input terminal for applying neuron input signals including a read portion and a write portion. The circuitry includes a read circuit producing a read signal dependent on resistance of the memory cell, and an output terminal providing a neuron output signal, dependent on the read signal in a first state of the memory cell. The circuitry also includes a storage circuit storing a measurement signal dependent on the read signal, and a switch set operable to supply the read signal to the storage circuit during application of the read portion of each neuron input signal to the memory cell, and, after application of the read portion, to apply the measurement signal in the apparatus to enable resetting of the memory cell to a second state.