Mixed Criticality Function Performance Optimization Based on Vehicle-Cloud Joint Scheduling

The optimization method of task scheduling has received increasingly extensive attention as the high level autonomous vehicle has been recognized as the typical mixed-criticality cyber-physical system. The low criticality functions are often degraded or abandoned at high system criticality level, thus decrease service satisfaction when the traditional mixed-criticality theory is applied to solve the scheduling problem. Under the scenario of the future prospect of intelligent transportation and the new electronic/electrical information architecture of network connection, this paper proposes an optimization method of vehicle-cloud joint scheduling, using edge computing to improve the performance of low criticality functions on the premise that high criticality functions can meet the deadline requirements. Since the previous DAG models used to describe vehicle functions are not specific enough, this paper also gives three detailed mixed-criticality function models to better fit the industrial reality. For the lack of evaluation of passengers’ perception when vehicle soft real-time DAG functions are not able to meet the deadline, the function completion evaluation coefficient is proposed to measure the scheduling effect of this situation, which also quantifies the degree of improvement of vehicle-cloud joint scheduling over the traditional local scheduling method.