Agent Spaces: a Scalable Architecture for Distributed Logic Agents

Paul Tarau, Arun Majumdar and Paulo Moura

October 2012


Abstract

We introduce a simple agent construct associated to a named local database and a Twitter-style weak inheritance mechanism between local agents.
On top of a remote predicate call layer, connecting distributed agent spaces, we build a replication mechanism allowing agents visiting remote spaces to expose their computational capabilities to non-local followers.
The resulting protocol has the remarkable property that only updates to the state of the agents are sent over the network through transactional remote predicate calls guaranteed to always terminate, and therefore spawning of multiple threads can be avoided. At the same time, calls to a visiting agent’s code by its followers are always locally executed, resulting in performance gains and reduced communication efforts.

Bibtex

@InProceedings{tarau-racs12,
  author =    {P. Tarau and A. Majumdar and P. Moura},
  title =     {{Agent Spaces: a Scalable Architecture for Distributed Logic Agents}},
  booktitle = {Proceedings of the 2012 Research in Applied Computation Symposium (RACS 2012)},
  pages =     {263--268},
  publisher = {ACM Digital Library},
  editor =    {M. Guo, J. Hong and E. Nadimi},
  month =     {October},
  year =      {2012},
  address =   {San Antonio, Texas, USA},
}

Download Paper

PDF file
ACM Digital Library