Probabilistic Inference Based Message-Passing For Resource Constrained DCOPs

Distributed constraint optimization (DCOP) is an important framework for coordinated multiagent decision making. We address a practically use-ful variant of DCOP, called resource-constrained DCOP (RC-DCOP), which takes into account agents ’ consumption of shared limited resources. We present a promising new class of algorithm for RC-DCOPs by translating the underlying co-ordination problem to probabilistic inference. Us-ing inference techniques such as expectation-maximization and convex optimization machinery, we develop a novel convergent message-passing al-gorithm for RC-DCOPs. Experiments on standard benchmarks show that our approach provides bet-ter quality than previous best DCOP algorithms and has much lower failure rate. Comparisons against an efficient centralized solver show that our ap-proach provides near-optimal solutions, and is sig-nificantly faster on larger instances.