A Scalable Architecture for Event-Based Cross-Correlation

Abstract

We present an architecture for a bio-inspired circuit that implements a wide-range cross-correlation. The circuit implements a stochastic approximation to mathematical cross-correlation. The basic circuit element is a noise-driven oscillator consisting of an integrator, a Schmitt inverter, and a switch which switches between the two signals to be cross-correlated. These elements can be developed into an NxN array which extends the cross-correlation range and improves the speed and accuracy of the approximation.