Input Variable Partitioning Method for Functional Decomposition of Functions Specified by Large Truth Tables

Abstract

Disk schedulers in current operating systems schedule a request as soon as the previous request has been completed. If the disk read requests are issued synchronously with short period between the scheduler switches to a request from another process with the assumption that the last process has no more requests. This condition is called deceptive idleness. It can be overcome by anticipatory disk scheduling framework, which introduces a short, controlled delay period, during which the disk scheduler waits for additional requests to arrive from the process that issued the last serviced request. Genetic algorithms, the most widely known types of evolutionary computation methods can be used for optimizing the anticipatory scheduling. In this paper, we propose to use the mutation operators while tuning the anticipatory scheduling with genetic algorithms. In general, a mutation operator is regarded as a secondary genetic operator but it plays a decisive role in the global functioning of the algorithm. We make the study on Linux operating systems kernel 2.6.16. We use various mutation operators and make a comparison on the performance of anticipatory scheduling.