Abstract

Mobile computing environments are characterised by significant and rapid changes in their supporting infrastructure and, in particular, in the quality-of-service (QoS) available from their underlying communications channels. Applications which can operate in these environments and take advantage of changing QoS require distributed systems support platforms. The current state-of-the-art in such platforms attempt to provide synchronous connection-oriented programming paradigms reflecting their fixed network origin. In this paper we argue that these paradigms are not well suited to operation in a mobile environment and instead propose a new platform called Limbo based on the tuple space communications paradigm. The design of Limbo is presented together with details of two prototype implementations. The use of the platform to re-engineer a number of existing adaptive mobile applications is also discussed.

Citations