Pseudo-Tree Construction Heuristics for DCOPs and Evaluations on the ns-2 Network Simulator

Abstract

Distributed Constraint Optimization Problems (DCOPs) are commonly used to model multi-agent coordination problems. However, empirical evaluations of DCOP algorithms are typically done in simulation under the assumption that the communication times between all pairs of agents are identical, which is unrealistic in many real-world applications. In this paper, we investigate the impact of empirically evaluating a DCOP algorithm under the assumption that communication times between pairs of agents can vary and propose the use of ns-2, a de-facto simulator used by the computer networking community, to simulate the communication times. Additionally, we introduce heuristics that exploit the non- uniform communication times to speed up DCOP algorithms that operate on pseudo-trees.