Solving The Multiple Constraint Satisfaction Problem (MCSP)

2363 Words5 Pages

Chapter 2 Related work 2.1 Introduction: Multiple constraint satisfaction problems (MCSP) is a problem in which a set of values must meet a number of constraints. it has been widely used in AI to solve a wide range of problems. In this thesis we use MCSP to solve the load balancing proble. Therefore, we first review the different algorithms in MCSP and the Dynamic MCSP then we review the load balancing problem. 2.2 Constraint Satisfaction Problem: CSPs are mathematical problems which are defined as a set of values that must meet a number of constraints or state restrictions. The problem is to search for a set of values for the features (or variables) so that the values satisfy some conditions (or constraints). A CSPs consists of a set of variables; For each variable, a limited range of possible values (domain); and a set of constraints that restrict the values of variables that can be taken at one time. CSPs solution is to assign a value for each variable, in such way that all assignment satisfy all restrictions or constraints. Since the possible values of the variables are limited, thus this kind of problem is combinatorial in nature and NP-complete. Formally, A CSP is defined as a triplet (X , D, C) where …show more content…

there is no consistent partial solution. it revises the variables by using hill-climbing (Makoto and Katsutoshi, 1996) search untill it reaches the ideal solution. The min-conflicts heuristic (Minton, et al., 1992) is a local search for solving CSPs. This heuristic chooses randomly variable in the scope of the restriction violated, and assigns it to a value in its domain that reduces the number of restrictions violated. If there is more than one value, it randomly selects among them. but the main weak point of the min-conflicts heuristic is that the possibility of being caught in a local, non solution minimum, which leads to restart the algorithm from a new initial

Open Document