Considering Equality on Distributed Constraint Optimization Problem for Resource Supply Network
Toshihiro Matsui, Hiroshi Matsuo · 2012 IEEE/WIC/ACM International Conferences on Web Intelligence and Intelligent Agent Technology · 2012
Distributed resource allocation is an important application of multiagent systems. In this work, we focus on a resource allocation problem that is motivated from a power supply network that contains distributed sources. In the supply network, resources that are initially distributed among source nodes have to be shared among all nodes. The problem is formalized as a resource constrained distributed constraint optimization problem that is an extended class of distributed constraint optimization problems (DCOPs). In the formalization, cost functions represent preferences of agents on resource use. We specifically consider allocating the cost values to agents as evenly as possible. We present several methods to select optimal assignment in consideration of the equality. The characteristics of the proposed methods are experimentally evaluated. Employing histograms and distinguishing types are effective for reducing the variance while requiring high computational costs. The number of histograms is able to be limited without significant lack of the effects. The minimum-maximum cost value as the main objective reduces the number of iterations and histograms.