Abstract
Distributed constraint optimization problem (DCOP) has attracted attention as one of the effective approaches for modeling distributed reasoning tasks in the multi-agent environment. However, existing approaches has alternatives problems i.e. requiring very long processing time to guarantee the optimal solution, or obtaining only a rough approximate solution by a probabilistic method at short time in the proposed algorithms. Under these constraints, existing approaches had issues when they are applied to the real world e.g. processing time issues in complete algorithms, and quality issues in approximate algorithms. In order to solve these problems, the authors had developed a method that uses Tabu Search under multi-agent environment, and we have improved this method by adding Simulated Annealing factor. This new method enables approximate solution under DCOP in short time. In this paper, we present the improved method adding Simulated Annealing factor, and report principle of algorithm and evaluation results comparing with existing algorithms, and application result to the large scale problem with 10000 variables.
