A Genetic Based Approach For Multiobjective Optimization Of Disk Scheduling To Reduce Completion Time And Missed Task

Abstract

Disk scheduling is a method by which a computer operating system decides the order that block I/O operations are submitted to storage volumes. It has an important role in QOS guarantee of soft real-time environments such as video-on- demand and multimedia servers. Many traditional algorithms exist to order the disk requests in an optimized manner. Many of these algorithms attempt to minimize any one parameter, completion-time or missed tasks at a time. In this paper, we propose a genetic based approach for disk scheduling which optimizes the completion time and number of missed tasks simultaneously. Traditional Multi-objective Genetic Algorithm(MOGA) is customized by including the concept of elitism to achieve better fitness values. The existing cycle crossover operator is modified to provide more search space. MOGA hybridized by combining with Simulated Annealing (SA) is known as Hybridized GA(HGA). In this method, the convergence of GA is improved by introducing the probability of SA as the criterion for acceptance of the new trial solutions. This hybridization shows more accuracy while finding solutions in later stage of searching process. The algorithms have been tested on several problems and the results were compared with traditional methods. Experimental results showed that the proposed GA based approaches worked very well and excelled most related works in terms of completion-time and number of missed tasks.