Analysis of Deterministic Online Scheduling

The relationship between the (open-loop) optimization problem that is repeatedly solved online and the quality of the (closed-loop) schedule that is implemented, is poorly understood, even in the deterministic case. We investigate various attributes of the open-loop problem and the rescheduling algorithm that affect the quality of closed-loop schedules, viz. rescheduling frequency, scheduling horizon length, and optimality gap. We find that it is beneficial to reschedule periodically even when there are no “trigger” events. Also, we show that solving the open-loop problem suboptimally does not necessarily lead to poor closed-loop solution due to the presence of feedback. Finally, we explore objective function modifications as well as addition of constraints to the open-loop problem as methods to improve closed-loop performance.