Scheduling with Finite Capacity Output Buffers

Abstract

In many scheduling problems, a job that completes processing may need to be held in an output buffer until the customer is ready to accept delivery. Buffer capacity is usually assumed to be infinite. We study a number of the best known single machine scheduling problems, under several alternative assumptions about the capacity of the output buffer. Specifically, we allow the buffer capacity to be either zero, fixed, or specified as part of problem input. We also consider situations in which all jobs have the same storage requirement in the buffer, and others where the storage requirement may vary. Further, we consider generalizations where there is a time interval within which a customer accepts delivery without cost to the producer. A classification scheme for these problems is provided. For each problem considered, we provide either an efficient algorithm or a proof that such an algorithm is unlikely to exist. Our results provide a mapping of the computational complexity of these problems which parallels those that are available for classical scheduling problems with infinite buffer capacity.