Publication

Abstract : Demand for Advanced Driver Assistance Systems (ADAS) is growing larger with the key focus of increased road safety, driver comfort and the futuristic autonomous vehicles. ADAS is typically a mixed-criticality system (MCS) with interdependent tasks which change their criticality level with the dynamic changes in the environment. In an MCS, the timing of safety-critical tasks must be strictly guaranteed to prevent accidents. From the conventional scheduling techniques in ADAS, more challenge is involved in meeting the timing performance of task scheduling and resource sharing while ADAS as MCS. This is a work-in-progress paper. Its objective is the design of ADAS with multicore partitioned architecture on the Xtratum hypervisor with the improved load and resource sharing among the subsystems. This is to address the dynamic changes in criticality and later to reduce the make-span for high critical tasks so as to prevent accidents. Two safety-critical ADAS subsystems Forward Collision Avoidance System (FCAS) and Blind Spot Avoidance System (BSAS) help to illustrate the architecture in this paper.