Back To the Future: IMS for LTE Can Learn from Class 5 Deployments

After arriving on the scene earlier this decade as the savior for networks, IMS has traveled a bumpy road in its evolution from its start as the solution for creating new applications over telco networks to an OPEX savings initiative, and finally the all-IP core for LTE networks (per the OneVoice initiative). While it has survived the hype cycle and settled into acceptance by most tier-one operators around the world, there are still different perspectives on IMS.  For some, this is still an initiative to launch innovative services while for others it is the core network that will bridge to legacy voice and eventually become the voice network.

While the business justification for IMS has adapted well with changing market dynamics, I’d argue that the deployment architecture has not evolved as much.  I’m not referring to the merged 3GPP/3GPP2/TISPAN architecture, which is just a “functional" representation of the elements: I mean real-world deployments. These still seem to follow a “data network" approach: a few very large, highly scalable boxes deployed in an operator’s major data centers in select locations around the country. That deployment model has, in turn, driven how the IMS core has for the most part been architected and tested – a few, very-high-capacity boxes that are centralized in the network.

That might work for an architecture that is used for data related applications and services, although it is not without its weaknesses. Large operators (a tier-one US wireless operator for instance), may typically have four or five large data centers around the country, even though they may have 50 or more MSCs (mobile switching centers) to handle voice in their geographically dispersed central offices. Chances are, if I make a call from my office in Westford, Mass., it is likely served by an MSC in nearby Woburn or Waltham – a radius of well under 50 miles.

For IMS to become the eventual core necessary to serve voice, we need to take a fresh look at the centralized, high-capacity design of IMS core networks. Sticking to a centralized approach ignores decades of learnings from the highly reliable PSTN network: Class 5 switches are many in number, local to a POP, and interconnected by a hierarchy of many fewer Class 4/Class 3 switches. A more efficient IMS network might be made up of smaller, self-sufficient network elements with the intelligence to serve voice and most other services locally, without requiring that signaling or media be backhauled to the center of the network. Services that require either special types of transcoding, or some pooled functionality, could be located at a few more centralized network elements.

Such a distributed approach is not only more fault resilient, it also is more capital efficient in that the service provider can grow its IMS footprint metro by metro, instead of committing to a huge infrastructure spend upfront. This is critical as significant upfront costs have been a barrier to large IMS network rollouts. The distributed approach also avoids the unnecessary backhaul that results from a centralized implementation. Just as with the rock-solid PSTN, IMS would benefit from moving intelligence to the borders of the network and leveraging all the smart network elements that can route, police or serve the ingress and egress traffic right where it touches a service provider’s network. 

Aashu Virmani, senior director of marketing at Sonus Networks, is an almost 20-year telecom industry veteran with extensive product and corporate marketing experience.  Prior to Sonus, Virmani served as a director of core network strategy for Motorola and lead marketing for mobile video infrastructure startup Aylus Networks.