Diffuse computing is an emerging paradigm in which computational tasks are performed by aggregated computational services, distributed over a network. This paradigm, developing rapidly as a result of commercial computing markets, the now-recognized potential of peer-to-peer systems, and the need for distributed network-centric systems, raises challenges for system design, software production, and the development of mechanisms ensuring stable equilibria of diffuse systems.
Diffuse computational systems require new software design and engineering methodologies. Components are combined on an as-needed basis, and with the increasing scale and complexity of modern distributed systems such as the WWW, the rigor and methodological support have failed to materialize in any form. With the local autonomy that characterizes large-scale distributed systems, global design and analysis may in fact be impossible.
This project proposes a synergistic approach to this problem, combining several complementary thrusts: incentive-compatibility in distributed computation, authorization mechanisms, secure data storage and retrieval, and communication protocols. This will include a formal treatment of underlying models, algorithms, and data structures. This research will be validated experimentally through evaluation of software prototypes.