QA-NoCs: Quantitative Analysis for Trojan Detection in Network-on-Chips
Padmaja Bhamidipati, Ranga R. Vemuri · 2024
Network-on-Chip (NoC) architectures have gained prominence in modern computing systems for their scalability and efficiency. However, the globalization of NoC design and fabrication exposes them to security threats. Quantitative analysis plays an important role in exposing vulnerabilities by quantifying packet information and traffic flow across components and pathways. It uses numerical data and mathematical models to understand complex systems, revealing patterns, and vulnerabilities invisible through qualitative methods. Our proposed QA-NoCs (Quantitative Analysis for Trojan Detection in Network-on-Chips) methodology leverages the power of quantitative analysis to protect NoCs against security threats. Our methodology dynamically evaluates network traffic to identify potential vulnerabilities. This novel lightweight approach introduces a quantifiable metric to identify traffic patterns, anomalies, attack scenarios, and potential information leakage points in NoCs. We evaluate our methodology on 4×4 and 8×8 mesh NoCs using real benchmarks. With Trojan infection rates as low as 1.5-2%, our approach shows an improvement of up to 3% in latency and 4% in throughput savings. Furthermore, our results show a substantial reduction in latency by over 50%, with a 2% to 10% Trojan infection scenario.