Scheduling objects in broadcast systems with energy-limited clients

Abstract

Broadcast systems are the popular infrastructure in push-based information distribution environments, where information objects are transmitted to subscribing clients, who are randomly switched on. The main problem in these systems is the construction of a periodic schedule (in cycles), where in every cycle each information object is transmitted several times; the number of appearances of each object is a function of its size and popularity. Existing algorithms try to transmit each object in a cycle with perfect periodicity, i.e. all object instances (appearances) in the cycle are equally spaced in time. These algorithms construct optimal schedules for environments with memory-less clients, optimizing the aggregate access delay for objects, and thus, minimizing client energy consumption accordingly. In this article, we provide an analysis of broadcast systems for memory-equipped (caching) clients. We change the scheduling optimization criterion to include actual object reception time in addition to access time, and thus, we provide a more realistic model for estimation of actual client power consumption. We prove that memory-equipped clients change the system model significantly, and allow for reduced object reception time, leading to improved energy consumption by clients. We give a simple proof of the fact that perfect periodicity in object transmission within scheduling cycles is necessary for optimal schedule, and calculate the conditions that optimal schedulers must satisfy. Since perfect periodicity is practically impossible to achieve (the problem is NP-hard), we analyze heuristic modifications of the broadcast schedule in order to achieve perfect periodicity for the more popular objects; heuristics include object transmission interleaving, preemptive transmission and exchange of object transmission order. We prove that interleaving should always be avoided and we calculate the conditions under which preemptive transmission and exchange of object transmission order result in reduced aggregate object reception delay, in systems with memory-less and memory-equipped clients.