Maximizing Parallelism without Exploding Deadlines in a Mixed Criticality Embedded System

International audienceComplex embedded systems today commonly involve a mix of real-time and best-effort applications. The recent emergence of low-cost multicore processors raises the possibility of running both kinds of applications on a single machine, with virtualization ensuring isolation. Nevertheless, memory contention can introduce other sources of delay, that can lead to missed deadlines. In this paper, we present a combined offline/online memory bandwidth monitoring approach. Our approach estimates and limits the impact of the memory contention incurred by the best-effort applications on the execution time of the real-time application. We show that our approach is compatible with the hardware counters provided by current small commodity multicore processors. Using our approach, the system designer can limit the overhead on the real-time application to under 5% of its expected execution time, while still enabling progress of the best-effort applications. I. INTRODUCTION In many embedded system domains, such as the automotive industry, it is necessary to run applications with different levels of criticality [13]. Some applications may have nearly hard real-time constraints, while others may need only best-effort access to the CPU and memory resources. A typical example is the car dashboard, which may display both critical real-time information, such as an alarm, and non critical information, such as travel maps and suggestions on how to outsmart traffic. Traditionally, multiple applications are integrated in a vehicle using a federated architecture: Every major function is implemented in a dedicated Electronic Control Unit (ECU) [28] that ensures fault isolation and error containment. This solution, however, multiplies the hardware cost, and, in an industry where every cent matters, is increasingly unacceptable. Recently, efforts have been made to develop an integrated architecture, in which multiple functions share a single ECU. AUTOSAR [16] is a consortium of actors from the automotive industry that defines a software architecture to exploit the benefits of integrated architectures by facilitating the reuse of applications. The AUTOSAR standard targets applications that control vehicle electrical systems and that are scheduled on a real-time operating system that is compliant with the AUTOSAR OS standard. Infotainment applications, however, typically target a Unix-like operating system, and thus still require the use of a federated architecture. Recent experimental small uniform memory access commodity multicore systems provide a potential path towards a complete low-cost integrated architecture. Systems such as the Freescale SABRE Lite [1] offer sufficient CPU power to run multiple applications on a single low-cost ECU. Using Virtualized architectures [8], [18], [34], multiple operatin