Dual Constraint Problem Optimization Using A Natural Approach: Genetic Algorithm and Simulated Annealing

Constraint optimization problems with multiple constraints and a large solution domain are NP hard and span almost all industries in a variety of applications. One such application is the optimization of resource scheduling in a pay per use grid environment. Charging for these resources based on demand is often referred to as Utility Computing, where resource providers lease computing power with varying costs based on processing speed. Consumers using this resource have time and cost constraints associated with each job they submit. Determining the optimal way to divide the job among the available resources with regard to the time and cost constraints is tasked to the Grid Resource Broker (GRB). The GRB must use an optimization algorithm that returns an accurate result in a timely mam1er. The Genetic Algorithm and the Simulated Annealing algorithm can both be used to achieve this goal, although Simulated Annealing outperforms the Genetic Algorithm for use by the GRB. Determining optimal values for the variables used in each algorithm is often achieved through trial and error, and success depends upon the solution domain of the problem. Although this work outlines a specific grid resource allocation application, the results can be applied to any optimization problem based on dual constraints