Resource Allocation for Covering Time Varying Demands

Abstract

We consider the problem of allocating resources to satisfy demand requirements varying over time. The input specifies a demand for each timeslot. Each resource is specified by a start-time, end-time, an associated cost and a capacity. A feasible solution is a multiset of resources such that at any point of time, the sum of the capacities offered by the resources is at least the demand requirement at that point of time. The goal is to minimize the total cost of the resources included in the solution. This problem arises naturally in many scenarios such as workforce management, sensor networks, cloud computing, energy management and distributed computing. We study this problem under the partial cover setting and the zero-one setting. In the former scenario, the input also includes a number k and the goal is to choose a minimum cost solution that satisfies the demand requirements of at least k timeslots. For this problem, we present a 16-approximation algorithm; we show that there exist well-structured near-optimal solutions and that such a solution can be found in polynomial time via dynamic programming. In the zero-one setting, a feasible solution is allowed to pick at most one copy of any resource. For this case, we present a 4-approximation algorithm; our algorithm uses a novel LP relaxation involving flow-cover inequalities.