https://doi.org/10.1140/epjs/s11734-025-01817-3
Regular Article
A fault-tolerant multi-value memory circuit design based on memristive neural networks
1
College of Electrical Engineering and Automation, Shandong University of Science and Technology, 266590, Qingdao, Shandong, China
2
College of Electronic and Information Engineering, Shandong University of Science and Technology, 266590, Qingdao, Shandong, China
3
Doctoral Workstation, Guangdong Songshan Polytechnic College, 512126, Shaoguan, Guangdong, China
Received:
6
March
2025
Accepted:
1
July
2025
Published online:
28
July
2025
Considering the limitations of CMOS technology in the nano-domain, memristors with non-volatile and nano-structures can replace CMOS for use in memory circuits and achieve multi-value synaptic weight storage in neural networks. However, the inherent defects and oxygen vacancy drift of the memristors may change the resistance and threshold voltage. Moreover, some memristors are insensitive to fixed pulses. Under these circumstances, the traditional fixed pulse scheme can not accurately complete the writing operation, which will degrade the reliability of multi-state weight encoding in memristive neural networks. To solve this problem, a multi-value memory circuit based on Au/Sr
Ba
TiO
/Pt memristors is designed for memristive neural networks. And a scheme based on feedback and the adjustment of signal amplitude is proposed by analyzing various faults of the memristor in the multi-value memory circuit. This scheme can enhance the stability and endurance of the memory circuit, improve the memory density, and reduce the interconnection complexity of the circuit. This paper provides a new insight into memory development in future.
Copyright comment Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature 2025
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
