Evaluation of communication models for community computing

Pervasive computing requires seamless communication among distributed software entities in heterogeneous environments. The community computing concept of PICO (Pervasive Information Community Organization) demands extensive communication and information sharing among software entities called delegents. The two modes of communication currently employed for distributed software entities are either direct or indirect. In direct communication, the sender explicitly addresses the receiver in order to interact, whereas in indirect communication, the interaction is through a third medium or shared space, sometimes called the blackboard. The objective of this work is to design, implement and evaluate communication framework for delegents in the service provider communities of PICO. Community interactions involve complex continual sharing of information among delegents at all times. Basic delegents and their communities are developed and deployed on a prototype distributed system employing state of the art tools. This thesis presents results of experimental work with two service provider communities, one with delegents using direct communication (message passing) and the other with delegents using indirect communication (space based communication). Results were obtained for different types of interactions involving information exchange between delegents of the community to determine the type of communication most suitable for different types of community interactions. Protocols for different community interaction are also proposed. An application of PICO was also designed, developed and tested for both the models to evaluate their performance in facilitating community computing.