Numerical Investigation on Structural Topology and Electrical Properties of Neuromorphic Nanowire Networks

The ability to understand the emergent dynamics within neuromorphic nanowire networks would significantly augment their applications and performance optimization. In this work, a numerical simulation approach combining graph theory with the HSPICE memristor circuit type model is proposed to capture and track the fundamental structural topology and electrical properties of the neuromorphic nanowire network. Numerical simulation results show that the increase in nanowire density can improve the structural topology metrics and the electrical features. In addition, the reduction of activation voltage can be related to the decrease of characteristic path length and the improved global efficiency. This may be attributed to the consequent decreased number of junctions in a percolative connectivity “winner-takes-all” path and the improved efficiency of the network in processing and transmitting current. This work provides further insights that help optimize neuromorphic nanowire networks in neuromorphic applications for achieving better performance.