Simulation Analysis of XOR Gates Implemented with a Memristor-Based Neural Network

We design and analyze a neuromorphic circuit that operates as a neural network and behaves as XOR logic gates. Artificial neurons are represented with operation amplifiers while weighted connections are realized with memristors. Generalized Verilog-A model is employed for memristor circuit simulations, which is calibrated on experimental data. We demonstrate control voltages used to read/write memristors and analyze its resistance levels and range. Additionally, power consumption simulation and Monte Carlo analysis are made to characterize XOR circuit dissipation and robustness to device-to-device variability, respectively.