A Hybrid Framework for Over-Constrained Generalized

Abstract

In this work we study an over-constrained scheduling problem where constraints cannot be relaxed. This problem originates from a local defense agency where activities to be scheduled are strongly ranked in a priority scheme determined by planners ahead of time and operational real-time demands require solutions to be available almost immediately. A hybrid framework is used which is composed of two levels. A high-level component explores different orderings of activities by priorities using Tabu Search or Genetic Algorithm heuristics, while in a low-level component, constraint programming and minimal critical sets are used to resolve conflicts. Realdata used to test the algorithm show that a larger number of high priority activities are scheduled when compared to a CP-based system used currently. Further tests were performed using randomly generated data and results compared with CPLEX. The approach provided in this paper offers a framework for problems where all constraints are treated as hard constraints and where conflict resolution is achieved only through the removal of variables rather than constraints.

Deep Space Network Scheduling via Mixed-Integer Linear Programming

“…The DSN receives many more requests than it can accomodate, which makes the completion time based objective functions commonly found in the scheduling literature lack meaning. Oversubscribed, overloaded, or overconstrained scheduling problems arise naturally in many domains such as airport gate assignment [24], processor task scheduling [25], [26], and meeting scheduling [27], [28]. Perhaps most similar to the DSN scheduling problem are problems in the satellite scheduling domain, including satellite range scheduling [29], [30], telescope scheduling [31], [32], payload scheduling [33], [34], and in particular the satellite observation scheduling problem [35]- [41], which also includes the time window constraints encountered in DSN scheduling.…”