Versatile CMOS Analog LIF Neuron for Memristor-Integrated Neuromorphic Circuits

Nikhil Garg,Davide Florini,Patrick Dufour,Eloir Muhr,Mathieu Faye,Marc Bocquet,Damien Querlioz,Yann Beilliard,Dominique Drouin,Fabien Alibart,Jean-Michel Portal

2024-06-28

Abstract:Heterogeneous systems with analog CMOS circuits integrated with nanoscale memristive devices enable efficient deployment of neural networks on neuromorphic hardware. CMOS Neuron with low footprint can emulate slow temporal dynamics by operating with extremely low current levels. Nevertheless, the current read from the memristive synapses can be higher by several orders of magnitude, and performing impedance matching between neurons and synapses is mandatory. In this paper, we implement an analog leaky integrate and fire (LIF) neuron with a voltage regulator and current attenuator for interfacing CMOS neurons with memristive synapses. In addition, the neuron design proposes a dual leakage that could enable the implementation of local learning rules such as voltage-dependent synaptic plasticity. We also propose a connection scheme to implement adaptive LIF neurons based on two-neuron interaction. The proposed circuits can be used to interface with a variety of synaptic devices and process signals of diverse temporal dynamics.

Emerging Technologies

What problem does this paper attempt to address?

The paper aims to address the following issues:

1. **Impedance Matching Problem**: In neuromorphic hardware, there is a significant impedance difference between CMOS neurons and memristor synapses, requiring impedance matching to achieve stable data reading.

2. **Low-Power Real-Time Simulation**: By developing a low-power characteristic analog Leaky Integrate-and-Fire (LIF) neuron circuit, it aims to achieve biologically realistic long-time scale dynamic simulation.

3. **Local Learning Rule Implementation**: A dual-leakage mechanism is proposed to support Voltage-Dependent Synaptic Plasticity (VDSP), thereby implementing local learning rules.

4. **Adaptive LIF Neuron Configuration**: A connection scheme is designed to enable generic LIF neurons to be converted into adaptive LIF neurons, enhancing information processing capability and memory capacity.

The paper addresses the above issues by implementing a circuit architecture that includes a Low Dropout Regulator (LDO), a current attenuator, and LIF neurons, and demonstrates its performance under different synaptic resistances, excitation voltage levels, and pulse widths. Additionally, the potential application value of this architecture in achieving efficient local learning is discussed.