Interest Reduction and PIT Minimization in Content Centric Networks

Abstract

Content Centric Networking aspires to a more efficient use of the Internet through in-path caching, multi-homing, and provisions for state maintenance and intelligent forwarding at the CCN routers. However, these benefits of CCN’s communication model come at the cost of large Pending Interest Table (PIT) sizes and Interest traffic overhead. Reducing PIT size is essential since larger memory sizes have an associated cost of slower access speeds, which would become a bottleneck in high speed networks. Similarly, Interest traffic may lead to upload capacity getting filled up which would be inefficient as well as problematic in case of traffics having bidirectional data transfers such as video conferencing. Our contribution in this paper is threefold. Firstly, we reduce PIT size by eliminating the need for maintaining PIT entries at all routers. We include the return path in the packets and maintain PIT entries at the egress routers only. Further, we use Persistent Interests (PIs), where one Interest suffices for retrieving multiple data segments, in order to reduce PIT entries at the egress routers as well as to reduce Interest overhead. This is especially useful for live and interactive traffic types where packet sizes are small leading to a large number of pipelined Interests at any given time. Lastly, since using PIs affects CCN’s original transport model, we address the affected aspects, namely congestion and flow control and multi path content retrieval. For our congestion scheme, we show that it achieves max-min fairness.