For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Scalable Cache Component in ICN Adaptable to Various Network Traffic Access Patterns
Atsushi OOKA Eum SUYONG Shingo ATA Masayuki MURATA
IEICE TRANSACTIONS on Communications
Publication Date: 2018/01/01
Online ISSN: 1745-1345
Type of Manuscript: Special Section PAPER (Special Section on Internet Technologies to Accelerate Smart Society)
information-centric networking, content-centric networking, caching, cache replacement algorithm,
Full Text: PDF(1.9MB)>>
Information-centric networking (ICN) has received increasing attention from all over the world. The novel aspects of ICN (e.g., the combination of caching, multicasting, and aggregating requests) is based on names that act as addresses for content. The communication with name has the potential to cope with the growing and complicating Internet technology, for example, Internet of Things, cloud computing, and a smart society. To realize ICN, router hardware must implement an innovative cache replacement algorithm that offers performance far superior to a simple policy-based algorithm while still operating with feasible computational and memory overhead. However, most previous studies on cache replacement policies in ICN have proposed policies that are too blunt to achieve significant performance improvement, such as first-in first-out (popularly, FIFO) and random policies, or impractical policies in a resource-restricted environment, such as least recently used (LRU). Thus, we propose CLOCK-Pro Using Switching Hash-tables (CUSH) as the suitable policy for network caching. CUSH can identify and keep popular content worth caching in a network environment. CUSH also employs CLOCK and hash-tables, which are low-overhead data structure, to satisfy the cost requirement. We numerically evaluate our proposed approach, showing that our proposal can achieve cache hits against the traffic traces that simple conventional algorithms hardly cause any hits.