Comparative study of A Cache Replacement Policies for Information-Centric Network

Abstract

Today's Internet user’s concentration on contents. Search engines, social media, play an important role (users pay an attention for contents which have a large number of like & share), ICN comes to meet the current internet user's needs. In ICN, every content is named, requesting data by name is very beneficial, especially for IOT. Now many researchers are studying ICN-based IoT systems. In this study, both router and user’s devices are able to store content. One of the key features of ICN is in-network content caching. ICN behavior is determined by a3-tuple, which are routing, content insertion and content replacement. Cache Placement strategies (e.g Caching everything-everywhere (CE2) stores all the content to be delivered on every router along the path, this approaches consumes greater amount of buffer space. As a result, a caching decision is made without regard to the content's popularity. for efficiency of caching, cache replacement policies play an important role, because the ICN router cache is limited, cache replacement is required. There are different cache replacement policies. This research compare between cache replacement algorithms (LUR, LFU, TLUR) in term of hit ratio to select the much more efficient one. Our goal is to find the optimal content replacement strategy, to face the big amount of caching (in-network caching) caused by placement algorithms, attempting to improve the network performance in terms of hit ratio and network delay, build model to compare algorithms. To calculate hit rate for every algorithm, and find the best algorithm for ICN. The application was built using web techniques (PHP, Javascript). The result of comparison explains that, the three algorithm is same when cache size is small, but the difference appears when the cache size become larger, whereas TLRU is the best one. LRU, LFU, TLRU achieves 12.5% with 4bytes cache size. While LRU, LFU achieves 25% hit, and TLRU achieves 29%hit with 8bytes cache size, using the same number of requests.