This paper presents a cache replacement policy which has specifically been developed for the efficient media caching in streaming media cache servers. For efficient media caching, the proposed policy takes into account the periodic patterns of users' requests in addition to the parameters such as reference count, amount of media contents delivered to the clients, and reference time. These values are collected at run-time for each cached object. In order to adequately and promptly adopt to the changing characteristics of users preferences, the policy introduces, in particular, the concept of weighted-window for replacement with which higher priorities are given to more recently referenced media contents and consequently they are less likely to be replaced. We present and analyze the simulation results showing that the proposed policy has outperformed the conventional replacement policies such as LRU, LFU, and SEG in terms of hit ratio, byte-hit ratio, delayed start, and cache input.
Bibliographical noteFunding Information:
This work was supported in part by the Basic Research Program of the Korea Science and Engineering Foundation (R01-2002-000-00141-0), 2004 Kwangwoon University Research Grant, and the ITRC programs (MMRC, HY-SDR) of IITA, Korea.
All Science Journal Classification (ASJC) codes
- Hardware and Architecture