Scalable information dissemination for pervasive systems: implementation and evaluation

Pervasive computing systems require large amounts of information to be available to devices in order to support context-aware applications. Information must be routed from the sensors that provide it to the applications that consume it in a timely fashion. However, the potential size and ad hoc nature of these environments makes the management of communications a non-trivial task. One proposed solution to this problem uses gossiping, a class of probabilistic routing protocol, to disseminate context information throughout the environment. Gossiping algorithms require far less in the way of guarantees about network structure, reliability, and latency than alternative approaches, but are unproven in real world scenarios. We describe the on-going development of a framework for evaluating the performance of these algorithms within the context of pervasive environments.