Performance-Driven Design of Engine Control Tasks

Engine control tasks include computational activities triggered at specific rotation angles of the crankshaft, making the computational load increase with the engine speed. To avoid overload at high speeds, simplified control implementations are used, defining different operational modes at different speed intervals. The design of a set of adaptive variable rate tasks is an optimization problem, consisting in determining the rotation speeds at which mode changes should occur to optimize the system performance while guaranteeing the schedulability. This paper presents three methods for tackling the optimization problem under a set of assumptions about the performance metric and the problem constraints. Two are heuristics and one is a branch and bound that is guaranteed, when it terminates, to find the optimum within a given granularity. In addition, a simple method to compute a performance upper bound is presented. The analysis of the problem reveals several insights for the design and the heuristics are shown to be quite close to the performance upper bound and the optimum with finite granularity.