Tuesday 17 June 2008

Flatter Architecture from Nokia-Siemens Network

From Unstrung:

In its bid to overtake Ericsson AB and become the world’s top radio access infrastructure supplier in terms of revenue, Nokia Siemens Networks believes its approach to all-IP flat architecture on 3G networks will give it an edge. Nokia Siemens says operators do not have to wait for LTE, to get the benefits of an all-IP architecture, and it is the only vendor that currently champions a flat 3G radio access network (RAN) approach.

As mobile data traffic continues to surge, operators are considering how to adopt flat, all-IP architectures in their 3G networks before the advent of 4G in order to gain lower latency, lower cost per bit, support for multiple access networks, and preparation for next-generation networks. But there are different ways to implement such architectures, and just how operators arrive at a flatter data network architecture is hotly debated.

Nokia Siemens has put its money on a flat RAN approach for high-speed packet access (HSPA) and the coming HSPA+ standard, in addition to its support for the Direct Tunnel architecture.
In a flat RAN architecture, the radio network controller (RNC) is integrated into the Node B so that the base station communicates directly with the Gateway GPRS Support Node (GGSN).
But there are as many benefits as drawbacks to flat 3G RANs, which makes it a controversial approach, according to the recent Heavy Reading report, "Flat IP Architectures in Mobile Networks: From 3G to LTE."

With flat RANs, some of the benefits include lower latency for data applications, lower operational costs due to fewer nodes to maintain and manage, augmented data capacity through a data network overlay, and good preparation for so-called 4G LTE/SAE (System Architecture Evolution), which uses a similar functional architecture. Also, costs won’t grow in line with data traffic growth, because operators won’t have to deploy extra RNC and SGSN capacity as traffic increases.

It may be challenging to integrate the RNC into a Node B. RNCs are critical to supporting macro-diversity in mobile networks, which enables mobile handsets to communicate with multiple base stations on the uplink and allows operators to deploy fewer base stations. NSN’s flat RAN architecture supports this feature, but in an unorthodox way, according to the Heavy Reading report.

So far, Nokia Siemens has three customers using its Internet HSPA (I-HSPA) flat RAN solution: Stelera Wireless and TerreStar Neworks in the U.S. and T-2 in Slovenia. And Mobilkom Austria AG & Co. KG recently trialed the solution.

Nokia Siemens’ Rouanne explains that flat 3G RANs aren’t necessary when there is just “medium” data traffic, but are best suited when operators have big data traffic volumes. “Those networks that are starting to be under pressure with traffic are coming to us and wanting to direct traffic directly to the Internet,” he says.

Even though Nokia Siemens is the only vocal supporter of flat 3G RANs right now, Brown says the strategy isn’t risky, but it’s “forward-looking.”

And a flat 3G RAN can set up an operator to be ready for the shift to LTE with its inherent flat architecture.

According to an old Ericsson presentation, ”Direct Tunnel” support added for 3G payload optimization has the following advantages:
  • Cost efficient scaling for Mobile Broadband deployments
  • Increased flexibility in terms of network topology
  • Allows the SGSN node to be optimized for control plane
  • Specifications part of 3GPP rel-7
  • Designed for operation in legacy (GGSN/UTRAN) networks
  • First step towards the SAE architecture
According to heavy reading article:
To efficiently deliver mobile broadband services, operators require a network infrastructure that simultaneously provides lower costs, lower latency, and greater flexibility. The key to achieving this goal is the adoption of flat, all-IP network architectures. With the shift to flat IP architectures, mobile operators can:
  • Reduce the number of network elements in the data path to lower operations costs and capital expenditure
  • Partially decouple the cost of delivering service from the volume of data transmitted to align infrastructure capabilities with emerging application requirements
  • Minimize system latency and enable applications with a lower tolerance for delay; upcoming latency enhancements on the radio link can also be fully realized
  • Evolve radio access and packet core networks independently of each other to a greater extent than in the past, creating greater flexibility in network planning and deployment
  • Develop a flexible core network that can serve as the basis for service innovation across both mobile and generic IP access networks
  • Create a platform that will enable mobile broadband operators to be competitive, from a price/performance perspective, with wired networks
Note: Diagrams above shamelessly copied from Ericsson's presentation.

No comments: