ALIFE: A multiagent computing paradigm for constraint satisfaction problems

This paper presents a new approach to solving N-queen problems, which involves a model of distributed autonomous agents with artificial life (ALIFE) and a method of representing N-queen constraints in an agent environment. The distributed agents locally interact with their living environment, i.e, a chessboard, and execute their reactive behaviors by applying their behavioral rules for randomized motion, least-conflict position searching, and cooperating with other agents, etc. The agent-based N-queen problem solving system evolves through selection and contest, in which some agents will die or be eaten if their moving strategies are less effective than others. The experimental results have shown that this system is capable of solving large-scale N-queen problems. This paper also provides a model of ALIFE agents for solving general CSPs.