Publication
IEEE Transactions on Knowledge and Data Engineering
Paper

Scheduling algorithms for the broadcast delivery of digital products

View publication

Abstract

We provide scheduling algorithms that attempt to maximize the profits of a broadcast-based electronic delivery service for digital products purchased, for example, at e-commerce sites on the World Wide Web. Examples of such products include multimedia objects such as CDs and DVDs. Other examples include software and, with increasing popularity, electronic books as well. We consider two separate alternatives, depending in part on the sophistication of the set-top box receiving the product at the customer end. The first, more restrictive option, assumes that the atomic unit of transmission of the product is the entire object, which must be transmitted in order from start to finish. We provide a solution based in part on a transportation problem formulation for this so-called noncyclic scheduling problem. The second alternative, which is less restrictive, assumes that the product may be transmitted cyclically in smaller segments, starting from an arbitrary point in the object. Three heuristics are provided for this difficult cyclic scheduling problem. Both scenarios assume that the broadcasts of the same digital product to multiple customers can be "batched." We examine the effectiveness of these algorithms via simulation experiments under varying parametric assumptions. Each of the three cyclic scheduling algorithms perform better than the noncyclic algorithm. Moreover, one of the cyclic scheduling algorithms emerges as the clear winner.