Impact Of Noise On The Performance Of Memristor

Memristor is a new basic two-terminal passive circuit element. It is a nanoscale device,exhibits non-linear current-voltage characteristics, and has an inherent memory property. In1971,Professor Chua Leon Ong predicted Memristor according to the symmetry principle. Thirty sevenyears later, i.e., in2008year, the scientists of Hewlett-Packard labs accidentally found memristorswhen they studied Nanostructured Titanium Dioxide Materials. They also proposed the physicalmodel of the corresponding memristor (HP model). Memristors can possibly allow for nano-scalelow power memory and distributed state storage, therefore it can be used for many applications,including digital memory, neuromorphic systems, and logic circuits. Many works focus on theworking principle and the corresponding physical properties of memristors based on the HP model.In practical engineering, noise always exists, such as the noise of the applied voltage and theinternal noise of the memristor. The presence of noise has an impact on the performance ofmemristor. At present, there are works focused on the influence of white noise, however thecorresponding discussions are not exhaustive. Few studies focused on other noises. We will studyextensively the impact of various noises on the performance of memristor.In this paper, we first introduce the physical model of memristor, its working principle and thefundamental electrical properties. Based on them, we analyze physical properties of memristor fordifferent initial parameters, which include the initial doping width, the maximum and minimumresistance ratio, and the frequency of the excitation voltage. The physical properties of thememristor can be characctered with functions of the doping width versus time, the current versustime, the current versus the voltage, and the flux versus the charge. Sinusoidal and triangle inputwaveforms are considered, respectively. We find that initial parameters will influence theperformance of memristor, so a perfect match of parameters should be considered when designing amemristor. Then we study the electrical characteristics of a single memristor under the Gaussianwhite noise, Gaussian color noise and pink color noise. Various ratios of signal to noise for the threetypes of noises are investigated. We find there are differences between the performances ofmemristor with noise and without noise. The memristor will lose its memory if the noise is strongerthan a threshold value, therefore in practice, a suitable memristor should be chosen according todifferent noise environments. At last, we study the impact of noise on the performance of twomemristors in series and in parallel, respectively. We find that if the noise is strong enough, it will destroy hysteresis curves and the circuit will lose its memory function.All our studies providevaluable design insights and allow a deeper understanding of key design implications ofmemristor-based memories.