Conditional and Multi-level WRITE Operations on Current-controlled Memristive Devices for Neuromorphic Applications
Abstract
The analog response of memristive devices makes them suitable for various applications, such as multilevel memory and neuromorphic computing. When the devices are voltage-driven, resistive switching is associated with voltage thresholds, representing the minimum required voltage to trigger a SET or RESET. Most simulation models correspond to voltage-controlled memristive devices, and multilevel tuning in such cases can be achieved via a controlled voltage-driven SET process. However, when current drivers are used instead of voltage drivers, a proper assessment of device performance should also consider models of current-controlled devices whose response adheres to current thresholds. To this end, here we provide both the voltage- and the current-controlled version of a behavioral model of memristive devices in a readily available netlist for simulation in LTSpice. We present simulation results using both model versions and show that, to achieve multilevel tuning of current-controlled devices when current-driven, a current divider circuit can enable a controlled RESET process. Finally, we provide experimental results for the controlled switching response of Self-Directed Channel (SDC) bipolar memristive devices by Knowm Inc. Overall, we demonstrate the feasibility of conditional and multilevel switching of current-controlled memristive devices, towards the design of current-based multilevel drivers and in-memory logic based on the conditional switching response triggered via current stimuli. © 2024 IEEE.
Más información
| Título según WOS: | Conditional and Multi-level WRITE Operations on Current-controlled Memristive Devices for Neuromorphic Applications |
| Título según SCOPUS: | Conditional and Multi-Level WRITE Operations on Current-Controlled Memristive Devices for Neuromorphic Applications |
| Título de la Revista: | Proceedings of the IEEE Conference on Nanotechnology |
| Editorial: | IEEE Computer Society |
| Fecha de publicación: | 2024 |
| Página de inicio: | 574 |
| Página final: | 579 |
| Idioma: | English |
| DOI: |
10.1109/NANO61778.2024.10628600 |
| Notas: | ISI, SCOPUS |