InterDigital looks at different protocols that can help smooth the delivery of video traffic
In my previous article we looked at the limitations of the traditional method of video delivery over the Web and particularly as it relates to mobile video. The primary issue was the scaling of demand, both in number of requests (users and requests per user) and file size (related to image quality), coupled with impending – and hard to reach – goals in terms of service delivery.
For the purposes of this article, we won’t look at codec- and compression-based approaches: my focus here is on network resources. And in delivering video, the focus can be on three areas: at the supply side (the streaming media company, for instance); at the demand side, with addressing and modeling user demand; and … somewhere in between.
One solution that has emerged for the time being is the content delivery network, which is an attempt to tackle the problem from the supply side. CDNs use domain name service level indirection and are an attempt to use scale of demand to reach some network-side efficiencies. If a viewer plays a video from high-demand sites and that demand can be predicted, the actual traffic and the actual video content can be predistributed to the CDN by placing the content at the operating side on specialized servers.
What complicates the CDN approach is security. It is important the video does not redirect to a site users do not trust via the CDN providing the video the viewer wanted. The redirection can look like an attack. Video delivery via this approach, in a secure and safe manner, is fairly complicated.
Tackling the problem from the demand side, another approach available by CDN providers like Akamai and others is HTTP proxy solutions. HTTP proxy solutions try to combine streams of personalized viewing into stable multicast groups. It is fairly complex to do and it’s not very suitable at a high aggregation level or very well suited for personalized end-user viewing due to the inherent personalized viewing by the end user. In other words, the lack of synchronization in both network characteristics as well as user behavior counters the attempt to form stable multicast groups, either forcing users into a synchronization and therefore behavior change, or leading to small multicast yield due to only few users fitting the pattern of aggregation.
But, in between the pure supply side and the user-focused demand side there’s a world of network architecture opportunity that can be explored, rethinking the equation. IP over IcN– the better IP is a European Commission-funded collaborative research and development project that uses intelligence coming out of information centric networking, a technology area being developed over the last eight to nine years. Said intelligence maps the HTTP traffic and HTTP level streaming onto a name-based model, which is what ICN is based on.
POINT translates protocols such as HTTP into something that is more efficient. The intelligence has the ability to do a very simple packet transfer in the networks that have done the mapping, and execute very simple binary operations in switches that are based on software-defined networking technologies. The trick is to do that mapping at the edge of the network. The solution allows for more efficient personalized viewing experiences by combining responses to similar viewing requests in order to optimize resources. For this, POINT’s approach essentially performs multicast at run time, which suits the randomized nature, and at the speed of packet forwarding making the delivery much faster and more efficient. In addition, POINT’s approach to latency reduction is the ability to dynamically relocate Web-based service resources, such as video play-out servers, and integrating these new resources at subsecond speeds, much faster than conventional DNS-based indirection methods.
The POINT approach also integrates very nicely with SDN by providing what’s called constant size flow tables. One of the issues in SDN as a technology is the flow tables explode in terms of size. This approach allows the flow tables to stay at a constant size, which is more secure and cost effective. So, it tackles a very serious problem at the moment with the way we integrate SDN into operational networks.
It is not clear yet which approach will take the lead. However, it is clear “5G” technology will be the catalyst that drives content providers to embrace new approaches to content delivery. Unlike previous generations of mobile technology – where more power was a suitable reaction to demand growth – 5G will require a significant rethinking.
Editor’s Note: In an attempt to broaden our interaction with our readers we have created this Reader Forum for those with something meaningful to say to the wireless industry. We want to keep this as open as possible, but we maintain some editorial control to keep it free of commercials or attacks. Please send along submissions for this section to our editors at: [email protected].
