Peer-to-peer networks for scalable grid landscapes in social agent simulations.

Craenen, B G W; Paechter, Ben

Abstract

Recently, peer-to-peer networks have been proposed as the underlying architecture of large scale distributed social agent simulations. A number of problems arise when grid landscapes are used to represent the landscape in these simulations, primarily because, in a peer-to-peer network, the landscape has
to be handled collectively by the nodes of the network. Two basic agent actions are identified as central to these problems: look and move. A solution to these problems is provided in which the network
maintains a move-buffer and a look-index. Both solutions are configurable by the user of the simulation and provide a trade-off between the scalability of the system, the consistency of the information stored in the system, and the efficiency of the system.

• Presentation Conference Type: Conference Paper (Published)
• Conference Name: AISB’05 Convention Social Intelligence and Interaction in Animals, Robots and Agents
• Start Date: Apr 12, 2005
• End Date: Apr 15, 2005
• Publication Date: 2005
• Deposit Date: Apr 28, 2010
• Publicly Available Date: Apr 28, 2010
• Peer Reviewed: Peer Reviewed
• Pages: 64-71
• Book Title: Proceedings of the Artificial Intelligence and Social Behaviour Convention (AISB) 2005
• ISBN: 1902956484
• Keywords: peer-to-peer networks; landscapes; simulations; virtual environments; move buffer; look index; scalability
• Public URL: http://researchrepository.napier.ac.uk/id/eprint/3395
• Contract Date: Apr 28, 2010