Analog-to-Stochastic Converter Using Magnetic Tunnel Junction Devices for Vision Chips

TL;DR

This paper presents a novel one-step analog-to-stochastic converter using magnetic tunnel junction devices, enabling area-efficient stochastic signal conversion for vision chips by leveraging the device's inherent probabilistic switching behavior.

Contribution

It introduces a one-step conversion method utilizing MTJ devices, reducing complexity and area compared to traditional two-step converters for stochastic computation.

Findings

Theoretical analysis of the conversion process.

Design and simulation of the converter in 90nm CMOS and 100nm MTJ technologies.

Consideration of resistance variability to improve conversion accuracy.

Abstract

This paper introduces an analog-to-stochastic converter using a magnetic tunnel junction (MTJ) device for vision chips based on stochastic computation. Stochastic computation has been recently exploited for area-efficient hardware implementation, such as low-density parity-check (LDPC) decoders and image processors. However, power-and-area hungry two-step (analog-to-digital and digital-to-stochastic) converters are required for the analog to stochastic signal conversion. To realize a one-step conversion, an MTJ device is used as it inherently exhibits a probabilistic switching behavior between two resistance states. Exploiting the device-based probabilistic behavior, analog signals can be directly and area-efficiently converted to stochastic signals to mitigate the signal-conversion overhead. The analog-to-stochastic signal conversion is theoretically described and the conversion characteristic is evaluated using device and circuit parameters. In addition, the resistance variability of the MTJ device is considered in order to compensate the variability effect on the signal conversion. Based on the theoretical analysis, the analog-to-stochastic converter is designed in 90nm CMOS and 100nm MTJ technologies and is verified using a SPICE simulator (NS-SPICE) that handles both transistors and MTJ devices.