Reliability-Aware Robust Scheduling for Heterogeneous Onboard High-Performance Computing

It includes a Python decision-level discrete-event simulator, test scripts, input parameters for an eight-task and four-resource heterogeneous computing scenario, a fixed random seed, CSV outputs for baseline comparison, disturbance-intensity sweeps, protection-budget sweeps, and independent resource-fault injection, together with figures generated from the source code.The simulation represents CPU, GPGPU, NPU, and FPGA resources.

It computes task-DAG scheduling decisions using resource compatibility, communication costs, first-order RC thermal dynamics, health-adjusted risk, and budgeted uncertainty protection. Static, HEFT, E-HEFT, NMC, and RaRS are evaluated with shared stochastic traces within each replay. The default configuration uses seed 20260915 and 2,000 replays per disturbance point.This package is a calibrated synthetic decision-level model.