Bloom Filter Implementation in Cache with Low Level of False Positive

Abstract

Searching techniques significantly determine the speed of getting the information or objects. Finding an object in a set is related to membership checking. In the case of massive data, it needs an appropriate technique to search an object accurately and faster. This research implements searching methods, namely Bloom Filter and Sequential Search algorithms, to find objects in a set of data. It aims to improve our system getting a proper item. Due to the possibility of False-Positive existence as a result of Bloom filter technique, there is a potentially inaccurate representation to object sought. Some parameters are influencing False-Positive, namely the number of objects, available bits, and the number of mappedbit. A Combination of those parameters could decrease the level of False-Positive and improve their accuracy and faster accessibility. In this research, we use three data object variations with the biggest object size of 2000000. Cached objects used in our experiments are between 2 -20% of variation from the generated objects. The best results with the lowest False-Positive is a combination of bit = 8, mapped bit = 7, and 6% of cache size from 2000000 generated objects.

Caching on Named Data Network: a Survey and Future Research

“…In paper [12] the package is deployed to be stored in network nodes but still maintained to ensure there are no redundant packets, to save resources, using bloom filter. The lack of bloom filter technique relatedto false positive problems corrected by A. Hidayat et al [13]. In this technique, bloom filter is combining with sequential search algorithm.…”

“…One time, a node can be a producer and sometimes it turns into theconsumer. In the NDN node can be embedded with various cache rules, including to determine which content will be selected and deleted from content store [5], [9], [10], the selection of places where a content will be cached [10]- [13], and cache policy that imply the model of cooperation between nodes to determinethe caching decision [14]- [17]. Related to node mobility, several techniques have been studied to maintain the performance of the system even though the nodes move in and out of coverage [18]- [21].…”