MCS-FT: A Fault-Tolerant AUTOSAR OS For Multicore Platforms

Abstract

The development of fault-tolerant systems in automotive applications, particularly under the AUTOSAR OS standards, faces significant challenges in managing mixed-criticality environments on multicore platforms. This paper introduces a novel fault-tolerant mixed-criticality system (MCS-FT) that employs advanced checkpointing and rollback mechanisms to ensure reliability and performance. A key innovation is the cooldown procedure, which dynamically decrements criticality levels, optimizes resource utilization, and reduces system overhead. This approach contrasts with traditional methods that focus solely on incrementing criticality levels. The MCS-FT model supports an unrestricted number of critical levels per core and allows user-configured resource-sharing protocols and fault management behaviors, constrained only by time for overrun and cooldown processes. Validation on the Infineon <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{Tricore}^{\mathrm{T}\mathrm{M}}$ TC $3\text{XX}$</tex> platform demonstrates substantial fault tolerance and system performance improvements, making MCS-FT a viable solution for safety-critical automotive applications. The proposed MCS-FT model meets and exceeds current standards, offering a new paradigm in designing and implementing fault-tolerant mixed-criticality systems.