Local Algorithms for Distributed Constraint Optimization in Dynamic , Anytime Environments

Many challenges in multi-agent coordination can be modeled as distributed constraint optimization problems (DCOPs) but existing complete algorithms do not scale well nor respond e ffectively to dynamic or anytime environments. As we move towards a future of ambient intelligence, in which large groups of agents must quickly coordinate, these deficiencies must be addressed. We decompose DCOP into a graphical game and investigate the evolution of various stochastic and deterministic algorithms. We also develop techniques that allow for coordinated negotiation while maintaining distributed control of variables. We prove monotonicity properties of certain approaches and detail arguments about equilibrium sets that o ffer insight into the tradeo ffs between e fficiency and solution quality. The ideas were tested in a simulated sensor network domain as well as a randomized, abstract domain.