Intelligent and predictive page replacement system
Proceedings of the Eighth International Multicon-W 2017: A Platform for Multiple Conference and Workshop
McGraw Hill Education (India) Private Limited
School of Science
In increasing the throughput of a system there are several things that matter. One of them is the page replacement algorithm. By managing the memory with least amounts of misses, the overhead is reduced thereby bringing down total time required in processing an application. An effective page replacement algorithm can also reduce the memory requirements and bring down the cost of the system. The best among them - Optimal Page Replacement Algorithm (OPT) - takes into consideration the near-future requirements of the pages and evicts pages that will not be required any sooner. The algorithm, however, requires that the future usage of the applications and pages be known. Innately, the future usage cannot be known in most cases, as each user uses his machine differently. This work is a bold attempt to overcome the problem faced in implementing the OPT algorithm, which is till now considered only theoretical, by using concepts of data mining and operating systems, and hence suggests a better page replacement algorithm. Simulations created for the suggested system and its performance analysis discussed in this work show significant promise. This work maintains a level of abstraction so that the algorithm can be scaled to any portable or fixed device (e.g. laptop, personal computer) and discusses the problems-solutions for each.
metadata only record