MULTILEVEL CONTENT STORE PERFROMANCE ANALYSIS ON NAMED DATA NETWORK

• Main Author: Putra Pamungkas, Muhammad
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43198
• Institution: Institut Teknologi Bandung

The internet is a technology that allows users to exchange information more flexibly. As internet service users increase, communication not only between human but also between machines (IoT), so does the need for internet access. With the architecture in use today, IP based network, Requires users to access server / producer (of the content) to obtain the desired content. This architecture will meet with limitations where the density of network traffic that will continue to increase. Named Data Network (NDN) is currently being developed as an internetsupporting architecture which changes the interconnection paradigm that was previously address-based to the desired content/information. NDN which is able to support the needs of users to access content through the content caching feature. The content caching capability allows the router to store copies of content in memory for a specified period of time. So users who need content can simply access the closest source (content store). The scheme that is widely used in the content caching method is to place content in one large memory. This research develops a scheme where cache memory is divided into two levels (multilevel CS). The purpose of using CS multilevel is to improve the performance of the content store. Testing is done by comparing the single content store scheme and multilevel content store. The parameters tested are the hit rate, miss rate, and average content lifetime in the content store. As a test parameter is a large change in content store size, interest rate, topology grid size, and consumer deployment. From the simulation results that the 2-level content store scheme has better performance compared to conventional schemes.