ARMOR-IMC: Adaptive Resource Mapping for Operational Robustness via Secure In-Memory Computing
2026-07-12 • Cryptography and Security
Cryptography and SecurityEmerging Technologies
AI summaryⓘ
The authors address problems in specialized computer chips that run AI models more efficiently by storing data inside the chip itself rather than moving it back and forth. They focus on two main risks: random manufacturing issues that cause errors, and security leaks from power consumption patterns. Their method improves the chip’s reliability and privacy after the AI model is already made, without needing to retrain it. Tests show their approach greatly fixes accuracy lost due to hardware errors and reduces the risk of attacks that spy on power use.
In-Memory ComputingDeep Neural NetworksSpin-Orbit Torque Magnetic Tunnel JunctionsProcess VariationSide-Channel AttacksFault InjectionPosttraining HardeningPower AnalysisSignal-to-Noise Ratio
Authors
Muhtasim Alam Chowdhury, Ramtin Zand, Soheil Salehi
Abstract
The massive data-movement overhead in traditional architectures has led to the adoption of In-Memory Computing (IMC) for energy-efficient Deep Neural Network (DNN) processing. By leveraging emerging devices like Spin-Orbit Torque Magnetic Tunnel Junctions (SOT-MTJs), IMC bypasses the "memory wall" and reduces leakage power inherent in traditional CMOS. However, this shift introduces dual hardware threats: manufacturing Process Variation (PV) degrades reliability and increases vulnerability to fault injection, while power Side-Channel Attacks (SCAs) compromise security. Existing defenses address these threats in isolation. This work presents a posttraining framework that simultaneously hardens analog IMC accelerators against both threats without retraining the model. Implemented in the IMAC-Sim simulator, our approach uses the proposed Variation Impact Score (VIS) to guide the mapping of Fault Observation Windows (FOWs) and introduces the Leakage Per Inference (LPI) metric to quantify input-dependent power variability under stochastic injection and the resulting reduction in effective signal-to-noise ratio. Experiments show that PV-induced faults can degrade accuracy by over 50%, while our method restores near-baseline accuracy and mitigates the threat of correlation-based power analysis attacks.