Pervasive Distributed Volume Rendering in a Lightweight Multi-agent Platform

In this paper, we present a service for a pervasive distribution of the computational load involved in the execution of high quality 3D rendering of medical volume data. The service is built upon an ad-hoc developed multi-agent middleware layer, which is in charge of executing user-defined tasks in hybrid peer-to-peer networks. Users' tasks are embedded in mobile agents and then spread in a platform composed by both fixed and mobile stations. Such agents are able to migrate among stations in order to ensure a certain degree of load balancing and dependability. The application scenario considered is a hospital, where intensive computing tasks can be e®ectively executed on a large number of heterogeneous devices that are networked and typically only partially used. We also present the results of performance and dependability tests performed under real-world conditions.