Publication Type : Conference Paper
Publisher : IEEE
Source : 2026 7th International Conference on Smart Systems and Inventive Technology (ICSSIT)
Url : https://doi.org/10.1109/icssit69151.2026.11656616
Campus : Amritapuri
Center : Humanitarian Technology (HuT) Labs
Year : 2026
Abstract : Hardware Trojans are malicious modifications in digital circuits, and they are not triggered until a specific unusual internal condition or input condition is fulfilled; therefore, they are difficult to detect through normal testing. The given paper presents a lightweight FPGA-based system that can be utilized to detect hardware Trojans at runtime with the help of an AI-driven monitoring system. An Arithmetic Logic Unit (ALU) with an embedded Trojan has been implemented on an FPGA, where the malicious code is triggered by a predefined runtime event and alters circuit behavior. The system measures outputs during operation and derives hardware activity such as signal transitions, switching patterns, and output changes. These features are fed into a lightweight Multilayer Perceptron (MLP) classifier developed using Vitis High-Level Synthesis (HLS), a tool that allows hardware design using C++ and automatically generates FPGA hardware code. The classifier learns to distinguish between normal and abnormal behavior and generates a Trojan alert in case of suspicious activity. This entire detection scheme operates in real time and does not require any external monitoring apparatus or offline analysis, making it suitable for secure embedded and reconfigurable systems. The experimental validation, conducted through simulation, demonstrates that the system can detect Trojan activation with low hardware cost, making it a practical and effective runtime security solution for FPGA-based digital designs.
Cite this Research Publication : Sreeraj Rajagopal Nair, Rajesh Kannan Megalingam, A Lightweight FPGA-based AI Framework to Detect Runtime Hardware Trojan, 2026 7th International Conference on Smart Systems and Inventive Technology (ICSSIT), IEEE, 2026, https://doi.org/10.1109/icssit69151.2026.11656616