Spin Torque Nano Oscillator based correlator

Abstract

Spintronics has emerged now as a viable post-CMOS technology. The main goal of this paper is to demonstrate pattern recognition using a cluster of eight Spin Transfer Torque (STT) devices biased to operate as variable frequency oscillators, Spin Torque Nano Oscillators (STNO). The phases of STNOs can be correlated to detect a trained input pattern. This is further used to design a correlator that detects the similarity of two patterns. While the spintronics devices are generally optimized for storing binary digital data, this paper shows how they can be configured to execute neuromorphic tasks without requiring any neurons and synaptic elements.