Simulated On-board AI-based Classification of Radiation-induced SRAM Event Upsets

This archive contains source code and data used to obtain the numericalresults, tables, and figures of the article "Simulated On-board AI-based Classification of Radiation-induced SRAM Event Upsets". Subject of the work: on-board classification of single-cell (SCU) andmultiple-cell (MCU) upsets in the BRAM memory of the AMD Artix-7 XC7A200T FPGA.Three configurations are compared: - Baseline - the original AICoRS system without on-chip classification; - Histogram - the offset-histogram method (accuracy reference); - MLP - the proposed neural-network classifier (8->32->16->1, Q8.8), deployed on the FPGA. The pipeline consists of three parts: (A) synthetic data generation and algorithmic analysis - MATLAB; (B) classifier training/evaluation - Python; (C) hardware implementation and resource/timing/power reports - Vivado. This archive contains parts (A) and (B) of the pipeline, including the datasets, analysis scripts, trained models, intermediate outputs, and the results used to generate the figures and tables reported in the paper.