Memristor-Based Neuron Simulation Circuit for Spatiotemporal Processing

The patent replaces traditional transistor-based computing units with neuron simulation circuits that use memristors to simulate biological neuron dynamics. The memristor-based circuits introduce ion dynamic processes that enable spatiotemporal information processing, substituting electronic transistor operations with electrochemical memristive behaviors that naturally capture temporal dynamics and ion channel effects of biological neurons

The patent changes the fundamental operating parameters of computing units from electronic transistor switching to memristive ion dynamics. By using memristors with adjustable conductance states and temporal response characteristics, the system achieves rich ion dynamic processes and spatiotemporal processing capabilities that transcend traditional binary transistor operations

The patent creates universal neuron simulation circuits that can perform multiple functions including integration, thresholding, spike generation, and temporal processing within a single circuit architecture. The memristor-based neuron circuits serve as multi-functional units that replace multiple specialized hardware components, reducing overall hardware resource requirements while maintaining spatiotemporal processing capabilities

The patent merges multiple neuron simulation circuits into array configurations where circuits are interconnected to form dense neural networks. By combining multiple memristor-based neuron units in compact arrays, the system achieves high computational density with reduced hardware resources per processing unit, enabling efficient spatiotemporal task processing