Timely Provisioning of Mobile Services in Critical Pervasive Environments

Abstract

Timeliness in conventional real-time systems is addressed by employing well-known scheduling techniques that guarantee the execution of a number of tasks within certain deadlines. However, these classical scheduling techniques do not take into account basic features that characterize today’s critical pervasive computing environments.

In this paper, we revisit the issue of timeliness in the context of pervasive computing environments. We propose a middleware service that addresses the timely provisioning of services, while taking into account both the mobility of the entities that constitute pervasive computing environments and the existence of multiple alternative entities, providing semantically compatible services. Specifically, we model the overall behavior of mobile entities in terms of the entities’ lifetime. The lifetime of an entity is the duration for which the entity is present and available to other entities. Given a new request coming from a mobile client and a number of semantically compatible mobile entities that can fulfill the request, one of them must be selected. The proposed service realizes three different policies that facilitate the selection. With respect to the first policy, the selection is realized solely on the basis of the client’s and the server’s lifetimes. The second policy additionally considers the load of each server towards selecting the one that guarantees to serve the new request within the lifetime of both the client and the server. The third policy further deals with periodic service requests.