Using Stochastic Computing To Defend Against Hardware Trojan Attacks

Description

Designed and implemented an alternative hardware-based Roberts edge detection implementation using stochastic computing in SystemVerilog that could withstand hardware trojan attacks. Designed and implemented an equivalent conventional binary implementation for comparison. Developed gate-level arithmetic modules and integrated hardware Trojan payloads and triggers to study the impact of malicious logic on image-processing pipelines. Built a full simulation and analysis workflow using Python and RTL simulation tools to evaluate accuracy and PSNR for stochastic and binary implementations under hardware trojan attacks. This project combined digital design, hardware security, and image processing to demonstrate how low-level circuit modifications can compromise system integrity while remaining difficult to detect.

Project Presentation

Presentation on stochastic computing hardware defense

Project Poster

Research poster

Project Report

Detailed project report

document