Unconventional arithmetic: A system for computation using action potentials

Abstract

This paper examines a scheme to perform arithmetic and logic computation using time delays inspired by neuronal Action Potentials. The method is reliant on a simple abstraction which utilises very little logical infrastructure, in fact, the only requirements necessary to carry out computation are a binary channel, a clock, and a rudimentary instruction look-up table. The conclusions are that the method is viable for all forms of arithmetic and logical computation including comparison, however one practical aspect that hinders a full move to a time delay based architecture is the inability to perform random memory access without waiting for the data to recirculate. © 2014 Springer International Publishing Switzerland.