Distributed constraint optimization for multiagent systems

• Authors: Wei-Min Shen;Milind Tambe;Pragnesh Jay Modi
• Venue: Distributed constraint optimization for multiagent systems
• Year: 2003

To coordinate effectively, multiple agents must reason about the interactions between their local decisions. Distributed planning and scheduling, distributed resource allocation and distributed task allocation are some examples of multiagent problems that require such reasoning. To represent these automated reasoning problems, researchers in Multiagent Systems have proposed constraints as a key paradigm. Previous research in Artificial Intelligence and Constraint Programming has shown that constraints are a convenient yet powerful way to represent reasoning problems. This dissertation advances the state-of-the-art in Multiagent Systems and Constraint Programming through three key innovations. First, this dissertation introduces a novel algorithm, named Adopt, for Distributed Constraint Optimization Problems (DCOP). Adopt is the first algorithm for DCOP that is asynchronous and guaranteed to terminate with the optimal solution. Adopt is empirically shown to yield orders of magnitude speedups over existing synchronous methods. The key idea is to perform distributed optimization based on conservative estimates rather than exact knowledge of global solution quality. Second, this dissertation introduces bounded-error approximation as a flexible method whereby agents can find global solutions that may not be optimal but are guaranteed to be within a given distance from optimal. This method is useful for time-limited domains because it decreases solution time and communication overhead. Bounded-error approximation is a significant departure from existing incomplete local methods, which rely exclusively on local information to obtain a decrease in solution time but at the cost of abandoning all theoretical guarantees on solution quality. Third, this dissertation presents generalized mapping strategies that allow a significant class of distributed resource allocation problem to be automatically represented via distributed constraints. These mapping strategies are significant because they not only illustrate the utility of our distributed constraint representation but also provide multiagent researchers with a reusable methodology for solving their own distributed resource allocation problems. These dissertation is intended for future researchers faced with solving distributed reasoning problems and opens the door for solving such problems under real-time dynamic conditions while providing theoretical guarantees on solution quality.