Core crunch. It’s not happening yet, but it might just be the catalyst to usher in 4G and force the industry to rethink where things are going in the wireless world.
With 3G reaching 336 Mbps (in Nokia-Siemens’ multi-band, multi-carrier, MIMO demonstration, which they say will be commercially available by the end of 2012) in an ecosystem that can easily co-exist with current 3G technologies, many question why we need LTE at all, which no longer offers the huge speed bump that once was its raison d’être. But there is still a compelling reason for LTE to modernize the backhaul with an all-IP core.
Laurent Persche, Asia-Pacific Business Development Manager at Alcatel-Lucent Velocix, believes that video will soon force networks to rethink the LTE vs 3G debate by pushing forward the need for intelligent IP-based content delivery caches in the LTE network.
With existing 3G, there is only one point in the network where there is pure, unencapsulated IP. After that, the data packets disappear into a shadowy world of circuit switched data and other non IP protocols. Today that model is still feasible, but think of a world with hundreds of thousands of users watching video and going about their daily online lives at 336Mbps and it quickly becomes clear that having one entry point for data quickly becomes untenable, and that an intelligent network with caching, much like the content delivery networks of the modern Internet, is needed.
Today, 70% of mobile traffic in the US is video, mainly Netflix, and the same is true in the UK for BBC iPlayer and the various video-on-demand services that terrestrial operators have adopted. Parts of Asia, such as India and Thailand are still very much broadcast markets, but China’s Youku already generates more traffic than YouTube worldwide. And City Telecom, parent of Hong Kong Broadband, has invested heavily in multimedia content.
Take for example multiple bit-rate H.264 videos. Many videos are streamed at 4 Mbps, 1 Mbps and often sometimes down to 360 Kbps. Add the Apple equivalents (they just have to do things their way) and it means that a typical video-on-demand episode needs seven different versions as people go from a uncongested network at home on a TV, to mobile while they travel, to when they get to the office pico or femto cell.
It means seven different versions need to enter the network at one point at the core in a 3G world.
With a modern CDN in an LTE network, it is possible to lay out an architecture for just one version to enter the network, for the content to be cached intelligently at nodes placed all around the network, and for the IP-aware CDN to generate lower bitrate versions of the video on the fly as radio congestion changes.
Core crunch will happen sooner than many think and that it will not just be uneconomical to run a 3G network, but technically impossible given the growth of data and expectations in our brave, new, wireless world.
Persche says the cut-off point will be at 25-MHz. A carrier with less than 25-MHz will face radio congestion first, but one with more will face a core crunch.
Femtocells will help delay the crunch, freeing up valuable spectrum by offloading to the Internet. The problem is that in many countries, such as Thailand, regulation is a grey area. Is a femtocell a consumer device? Or is it a base station? If it is a base station, then there are costs and ownership issues in setting up a new base station as the telco needs to pay per base station and transfer ownership to the concession holder.
In the 3G world, the closest a cache can get while remaining IP is when it enters the Telco. With LTE, caching nodes can be placed anywhere in the network, using existing tried and tested technology, making video delivery much smoother, much faster and much cheaper.
But some argue that 3G can do data too and caching and content delivery can be done in a 3G context without having to go to LTE. Persche says that the argument is similar to those saying that IPv6 is not needed as everyone can use NAT. There are just too many compromises and issues that need to be worked out for that to happen.