Scalable and self-organizing (meta) data dissemination for ubiquitous computing

One of the promises of ubiquitous computing is the ability to tie heterogeneous computing elements together in a manner where the whole is larger than the sum of the individual parts. Thus, Internet services, file servers (PCs), personal digital assistants, sensors and embedded computers (in appliances ranging from light switches to automobiles) interact with one another based on application requirements and proximity. One key challenge to coordinating computers in the above manner is the scalable dissemination of service data and meta-data to a potentially large population of clients. We discuss a number of examples. The common thread running through all these applications is the need for disseminating data in a scalable fashion among potentially millions of nodes spread across the Internet. We next present Dicast, a communication model designed to support scalable dissemination for a range of applications and semantics. We then discuss a number of research issues associated with realizing this model.