Optimizing for Video Storage Networking With Recommender Systems
Bell Labs Technical Journal
| Hi-index | 0.00 |
Of all the services offered over a worldwide public network, digital video-based services, even at this early stage of development, are growing at such an astonishing rate that they will soon constitute the largest portion of Internet traffic. We are witnessing this today with the Internet television (TV) phenomenon, with YouTube, Netflix's online offering, and BBC's iPlayer serving as prominent examples. To optimize video-dominated networks with respect to the required transport capacity (and also to reduce the response time the end user experiences), these services need to be supported by new networking capabilities. Caching in conjunction with specific unicast and multicast techniques for ingress and egress traffic can achieve this goal. In this paper we discuss how, explicitly by using cache cooperation techniques, we can deal with heterogeneous content preferences among end users. The paper shows simulation results and discusses relevant parameters with respect to the Internet TV use cases mentioned. Finally, it provides an outlook on the longer term and demonstrates that investment in these new capabilities is future-safe. © 2009 Alcatel-Lucent.