Mimicking spike-timing-dependent plasticity with emulated memristors

Abstract

Memristors have found application in neuromorphic circuits and it has been shown that, under certain conditions, they may mimic the behavior of neuronal synapses. Experimenting with memristor-based synapses has several problems. Indeed, memristor samples are difficult to obtain and tweaking their parameters to adapt their behavior requires a long fabrication and testing process. For this reason, simulation and emulation become attractive alternatives for the study of memristive systems. We postulate that emulation has the advantage of working with real-world circuits and not stylized simulation models. In this paper, we propose a basic memristor emulation architecture and show that it can be used to mimic certain characteristics of synapses.