The patent describes a Partially Depleted (PD) Silicon-On-Insulator (SOI) MOSFET designed to emulate Leaky Integrate-and-Fire (LIF) neuron behavior. The device includes a substrate, a buried oxide layer, a channel region, source and drain regions, a gate oxide layer, a gate, and a reset circuit. The MOSFET mimics the neuron's spiking behavior through a series of electronic processes, including impact ionization and hole storage.
Figure 1. (a) Device schematic with biased terminals, (b) Input biasing scheme (VSG(t),VDG(t)) with expected output (ID(t)), (c-i) Equilibrium band diagram. (ii) Electron hole pair generation due to impact ionization (
Traditional implementations of neuron-like circuits either lack area efficiency, are not CMOS compatible, or are power inefficient. Solution: This invention addresses these issues by utilizing a PD SOI MOSFET, which is CMOS compatible and incorporates an efficient reset circuit and impact ionization mechanism to emulate neuron behavior.
Salient technical feature(s):
- PD SOI MOSFET structure.
- Impact ionization enabled the floating body effect.
- Efficient reset circuit with capacitors for RC time-constant.
- CMOS compatibility.
Advantages of the technology:
- Area efficiency
- CMOS compatibility
- Power efficiency
- Simplified integration into existing semiconductor processes
The patent describes the structure and operational method of the PD SOI MOSFET configured to emulate LIF neuron behavior.
This invention is in its stage of testing and evaluation stage for the final objective.
3
This technology could revolutionize artificial intelligence and neurocomputing by providing more efficient and scalable hardware for mimicking neuronal behavior, potentially leading to advancements in brain-inspired computing and smart systems.
- Neuromorphic computing
- Artificial neural networks
- Brain-inspired computing systems
- Advanced computational devices
Geography of IP
Type of IP
201721027169
525199