A Virtualized Architecture for Software-in-the-Loop Testing Applied to the LEON3 Processor

Abstract

The increasing complexity of embedded systems in aerospace missions, particularly within the New Space paradigm, calls for more agile and cost-effective approaches to software and hardware integration. Traditional prototype-heavy development cycles are being replaced by virtualization and emulation techniques that support faster, iterative validation. Despite the growing adoption of such techniques, few studies propose a flexible and stable software-in-the-loop (SIL) framework tailored to emulated environments in the aerospace sector, especially considering open-source and widespread tools and technologies. This work addresses this gap by introducing a formal and adaptable SIL testing architecture based on the Quick EMUlator (QEMU), an open-source emulation platform, as its core. The framework targets the LEON3 processor, widely used in aerospace applications, and was validated through three sequential test scenarios integrating emulated environments and physical counterparts. These tests assessed software correctness, logical consistency, and timing behavior. Results confirmed full test success rates and the logical fidelity of the virtualized system, while revealing inherent timing discrepancies, characterized by an average advance of 20 ms in processing and transmission times compared to the physical counterpart. Despite these differences, the framework demonstrated sufficient accuracy and reliability for software testing in virtualized environments, provided its timing variations are properly accounted for.