Self-organizing approaches for large-scale spray multiagent systems

Large-scale multiagent systems will be the key software technology driving several future application scenarios. We envision a future in which clouds of microcomputers can be sprayed in an environment to provide, by spontaneously networking with each other, an endlessly range of futuristic applications. Beside this vision, similar kind of large-scale "spray" multiagent systems will be employed in several other scenarios ranging from ad-hoc networks of embedded and mobile devices to worldwide distributed computing. All of these scenarios present strong commonalities from the application development point of view, and new approaches and methodologies will be likely to apply, to some extent, to all of them. In particular, we argue that the issues related to the design and development of such spray multiagent systems call for novel approaches exploiting self-organization as first-class tools. With this regard, we survey a number of research projects around the world trying to apply self-organization to large-scale multiagent systems. Finally, we attempt at defining a rough research agenda that-in the long run-should integrate these ideas to develop a general and more assessed methodology for large-scale spray multiagent systems crosscutting several application domains.