Chemically-inspired memristor-based neuron-like oscillating circuit

Chatzinikolaou, Theodoros Panagiotis; Fyrigos, Iosif Angelos; Tsioustas, Charalampos; Bousoulas, Panagiotis; Tsompanas, Michail Antisthenis; Tsoukalas, Dimitris; Sirakoulis, Georgios Ch

Abstract

There is a constant push for ever increasing performance in traditional computing systems, leading to high power consumption and, in the end, to the incapacity of conventional electronics to handle heavy computing tasks, which usually require learning features. Thus, the development of novel nano-electronic devices with inherent neuromorphic characteristics and a low energy footprint has become a viable alternative. In order to simulate neuromorphic features utilizing memristive devices, the threshold switching effect is critical, which can be seen in the rich dynamics of metallic conductive filament (CF). In this paper, a realistic model of the unipolar nature of CBRAM devices is exploited to create a memristor-based oscillator that can integrate neuromorphic features. Bipolar memristive devices have been used to match the weight of the neurons in a crossbar configuration. The used physical model for these memristors was fitted to fabricated devices in order to achieve the expected accuracy in the circuit simulation. The oscillator's output signal and behavior matched the theoretical background of biological neurons. Thus, this approach can be considered as the first step towards the development of low-power oscillation-based neuromorphic hardware with biological-like behavior.

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2022 Panhellenic Conference on Electronics and Telecommunications, PACET 2022
• Start Date: Dec 2, 2022
• End Date: Dec 3, 2022
• Acceptance Date: Sep 30, 2022
• Online Publication Date: Dec 14, 2022
• Publication Date: Dec 14, 2022
• Deposit Date: Jun 2, 2024
• Publisher: Institute of Electrical and Electronics Engineers (IEEE)
• Book Title: 2022 Panhellenic Conference on Electronics & Telecommunications (PACET)
• ISBN: 9798350399592
• DOI: https://doi.org/10.1109/PACET56979.2022.9976319
• Public URL: https://uwe-repository.worktribe.com/output/12030994