Performance Evaluation of the Replacement Policies for Pending Interest Table

Abstract

Information (content) plays an important role in a Named Data Networking (NDN). Hence, an information model is essential in representing information appropriately to supports meaningful information spreading. As a distinction from the current network practice, the NDN shall concentrate on the content itself, rather than the location of the information itself. One important and common feature of NDN is leveraging through its built-in network caches (temporal store) to improve the communication and efficiency of content dissemination. Thus, caching is well thought-out as one of the most crucial features (especially in PIT) of the NDN. Its efficiency is due to it required feature of producing a flexible strategy in deciding what content to store and replace when the PIT overflows. Thus, PIT management in NDN continues being one of the primary concerns of high-speed forwarding. To address this issue, replacement policies, as one of the key factors for determining the effectiveness of a PIT in line with many researcher's haven to propose numerous replacement policies, i.e. LRU, Random and Persistent, which have been projected to attain the improved Interest drop rate, reduce the delay and Interest retransmission as when the PIT is full. However, to the best of our knowledge, there have not been studies that dealt with the performance and evaluation between the mentioned policies under different network topologies. Therefore, in this paper we study the performance of Interest drop rate, delay and Interest retransmission under different network topologies, i.e. Tree, Abilene and Germany when the PIT is full. The significance yearned for this study would be to provide a solid starting point in research directions of new PIT replacement policies for contemporary workload or selectively turning off of fewer used cache ways.