M. Saito and M. Noto (Japan)
multi-agent, resource allocation, distributed constraint optimization, asynchronous distributed optimization, localload distribution
Distributed constraint optimization is very important for solving asynchronous distributed search problems in the multi-agent field. Various methods have been proposed for solving distributed constraint optimization problems, and among these, Asynchronous Distributed OPTimization (ADOPT) has proven to be very effective. ADOPT is able to find solutions that optimize the total cost, but depending on the distribution of solutions for the problem in question, loads can tend to localize. When handling practical problems, solutions that take local load into consideration may be required. If this type of load concentration can be resolved, it should be possible to apply this approach to various resource allocation problems such as frequency allocation or sensor network allocation. In this paper, we propose a method that improves on ADOPT by taking distribution of local loads into consideration, making it applicable to these sorts of practical problems. We also verify the effectiveness of this new method by performing simulation experiments using the frequency allocation problem.
Important Links:
Go Back
