A new multi-agent algorithm for solving constraint satisfaction problems

Constraint satisfaction problem is considered as an important area of research in artificial intelligence, so that many problems in artificial intelligence and other fields of computer science can be defined as constraint satisfaction problems (CSPs). On the other hand, there are situations where a problem needs to be solved in a distributed mode, such that it is impossible to use a central controller. In these conditions, the importance of multiagent systems is specified. In this paper we will present a new multi-agent method called CSC (constraint satisfaction community) to solve those CSPs which have Alldiff constraints, namely, those issues which all variables should have different values. Experimental results show that this algorithm achieves good performance when the size of problem increases, and almost has a linear time complexity.