Software-defined content delivery network at the edge for adaptive video streaming.

In DASH (Dynamic Adaptive Streaming over HTTP), it is required to store multiple encoding for same video segments which increases the storage requirement of CDN (content delivery network) servers. Further, to keep network delay in check, it is not feasible to continuously increase the number of CDN servers in different geographical locations as per the ever-increasing demand. Peer-to-peer (P2P) CDN acts as an elegant alternative to supplement existing CDN servers to harness the benefits of both P2P and CDN technologies. We present the concept of "CDN at the edge," where end hosts can collaborate for video streaming in pure P2P manner to provide better QoE (Quality of Experience). However, dynamically changing behavior of underlying network, uncertain leaving, joining pattern of clients, and frequent change in source node during video streaming session make the selection of the desired bitrate (at any given time) extremely difficult. In this paper, we have designed and implemented a software-defined network (SDN)-based P2P CDN system that supports browser-based adaptive video streaming. SDN provides global view of the topology and thus assists in providing timely information about leaving or joining peers in the proposed "CDN at the edge." Further, this SDN-based P2P CDN is compared with non-SDN-based P2P CDN using both virtual switch-based emulations and physical switch-based actual implementation. The comparison is performed under different churn rates with respect to QoE parameters such as start-up delay, bitrate, stall count, and its duration. The experimental results prove the applicability of our approach.