The improved local cost simulation algorithms based on coalition structure generation for solving distributed constraint optimization problems

Distributed constraint optimization problems (DCOPs) are a fundamental framework for modeling multi-agent systems (MAS) where agents coordinate their decision-making to optimize a global objective. However, existing incomplete local search algorithms for solving DCOPs face a huge challenge of fallin...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:The Journal of supercomputing Jg. 81; H. 1; S. 132
Hauptverfasser: Shi, Meifeng, Jia, Guoyan, Yokoo, Makoto
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York Springer US 01.01.2025
Springer Nature B.V
Schlagworte:
Algorithms
Asymmetry
Compilers
Computer Science
Constraints
Design
Interpreters
Multiagent systems
Optimization
Optimization techniques
Processor Architectures
Programming Languages
Search algorithms
Simulation
Solution space
Variables
Local search
Coalition structure generation
Symmetrically multiple neighbor ignoring
Distributed constraint optimization problems
ISSN:0920-8542, 1573-0484
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Distributed constraint optimization problems (DCOPs) are a fundamental framework for modeling multi-agent systems (MAS) where agents coordinate their decision-making to optimize a global objective. However, existing incomplete local search algorithms for solving DCOPs face a huge challenge of falling into the local optimum since the information and controls are distributed among multiple autonomous agents. Considering this, two improved local cost simulation algorithms based on coalition structure generation (CSG) named LCS-CSG-F and LCS-CSG-C that execute CSG in fixed and consecutive rounds, respectively, are proposed in this paper to expand the search for solution space. CSG involves partitioning a set of agents into disjoint exhaustive coalitions according to the neighbor relationship between agents, where agents of the original DCOPs are partitioned into coalitions to relax constraints between agents. To verify the effectiveness of the CSG strategy, two competing schemes that execute neighbor ignoring in fixed rounds, named asymmetric single neighbor ignoring (ASI-F) and asymmetric multiple neighbor ignoring (AMI-F), are designed since the CSG strategy symmetrically ignores multiple neighbors in different coalitions simultaneously. From statistical perspective, the Friedman test reveals significant differences between the LCS-CSG algorithm and other algorithms. Extensive experimental results indicate that the proposed CSG-based algorithms significantly outperform the state-of-the-art DCOPs incomplete algorithms in various benchmarks.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0920-8542
1573-0484
DOI:10.1007/s11227-024-06644-2