A Rank Based Replacement Policy for Multimedia Server Cache Using Zipf-Like Law

TL;DR

This paper introduces a novel rank-based cache replacement policy for multimedia servers that leverages Zipf-like law to improve cache performance by considering video popularity, size, and access cost.

Contribution

The paper proposes a new cache replacement algorithm that incorporates video characteristics and Zipf-like law for ranking, outperforming traditional methods like LRU and LFU.

Findings

Improved cache hit ratio and cache byte ratio

Reduced average request latency

Outperforms LRU, LFU, and Greedy Dual algorithms

Abstract

The cache replacement algorithm plays an important role in the overall performance of Proxy-Server system. In this paper we have proposed VoD cache memory replacement algorithm for a multimedia server system. We propose a Rank based cache replacement policy to manage the cache space in individual proxy server cache. Proposed replacement strategy incorporates in a simple way the most important characteristics of the video and its accesses such as its size, access frequency, recentness of the last access and the cost incurred while transferring the requested video from the server to the proxy. We compare our algorithm with some popular cache replacement algorithm using simulation. The video objects are ranked based on the access trend by considering the factors such as size, frequency and cost. Many studies have demonstrated that Zipf's-like law can govern many features of the VoD and is used to describe the popularity of the video. In this paper, we have designed a model, which ranks the video on the basis of its popularity using the Zipf-like law. The video with higher ranking is named "hot", while the video with lower ranking is named "cold". The result show that the proposed rank based algorithm improves cache hit ratio, cache byte ratio and average request latencies compared to other algorithms. Our experimental results indicate that Rank based cache replacement algorithm outperforms LRU, LFU and Greedy Dual.