Abstract
The Distributed Constraint Optimization Problem (DCOP) is a powerful framework for modeling and solving applications in multi-agent coordination. Asynchronous Forward Bounding (AFB-BJ) is one of the best algorithms to solve DCOPs.We propose AFB-BJ+, a revisited version of AFB-BJ in which we refine the lower bound computations. We also propose to compute lower bounds for the whole domain of the last assigned agent instead of only doing this for its current assignment. This reduces both the number of messages needed and the time future agents remain idle. In addition, these lower bounds can be used as a value ordering heuristic in AFB-BJ+. The experimental evaluation on standard benchmark problems shows the efficiency of AFB-BJ+ compared to other algorithms for DCOPs. © 2013 Springer-Verlag.
Cite
CITATION STYLE
Wahbi, M., Ezzahir, R., & Bessiere, C. (2013). Asynchronous forward bounding revisited. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8124 LNCS, pp. 708–723). https://doi.org/10.1007/978-3-642-40627-0_52
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
