An adaptive neuro fuzzy inference system for cache replacement in multimedia operating system

In this paper, we propose an adaptive cache replacement scheme based on Adaptive Neuro Fuzzy Inference System (ANFIS), as a part of virtual memory management system of a multimedia operating system. Existing general purpose cache replacement mechanisms have been developed for a single class of applications. Since these mechanisms treat all applications alike regardless of their requirements, they are ineffective at simultaneously supporting multiple application classes. Since none of the replacement policy is best, it can be observed form the studies that the disadvantage of one replacement policy can be overcome by another replacement policy. But the key lies in the decision for selecting the best policy. We have proposed a system that implements number of policies and dynamically adapts to one of them depending upon the class of application. The proposed system decides the best policy using ANFIS. The system includes ANFIS model that implements number of policies and determine the performance of each policy for a particular application in terms of several performance criteria. Simulation results reveal that our scheme can provide remarkable improvement in the miss ratio as compared to conventional algorithms. We demonstrate that our scheme outperforms LRU which is considered as the best replacement policy.