Monday, December 15, 2008

ST-NXP sees "strong demand" for 3G/UMA

In an article released today at EE Times Europe, Dan Rabinovitsj, senior vice president and general manager of cellular systems for ST-NXP said the company sees strong demand from device manufacturers and operators for 3G/UMA devices.

ST-NXP, the new combined entity of ST Microelectronics and NXP which merged in April, 2008, brings a lot of strength in UMA technology with NXP, an early innovator and supporter of UMA.

Mr. Rabinovitsj continues:

"With the Cellular System Solution 7210 UMA, we are able to deliver high-quality streaming video and fast download speeds. In a Wi-Fi hotspot, consumers will notice the difference in speed when accessing popular social networking sites, and the smooth video quality when watching their favorite clips online. This brings to consumers the broadband speed they typically enjoy on the PC, through a UMTS feature phone."

We have already seen a number of 3G/UMA phones on the market, and it appears that there are many more on the horizon.

UMA, and Wi-Fi technology, is an ideal complementary technology to 3G. Both deliver high-speed broadband access for mobile services, while UMA/Wi-Fi can augment 3G coverage in the places where consumers spend the most of their time, at home and in the office.


Anonymous said...

What is the benefit to the end users to use UMA over WiFi to access data services? Wouldn't it be more efficient to access the internet/data services directly over WiFi access?

Steve said...

Excellent question/comment. The short answers are ‘none’ and ‘yes’, and here’s an explanation of how/why UMA supports local traffic offload.

There are two types of data services on a mobile phone: those which come from the mobile operator (ie Orange’s HD TV service) and those which are simply accessing Web 2.0 services on the Internet (YouTube, Google,…)

For the first case, it’s pretty clear that the mobile operator benefits by offloading mobile packet data services with UMA. UMA is cheaper and faster than the outdoor macro network. Note that in this case, Wi-Fi alone is no help. Wi-Fi without UMA can not access any of the services located inside the mobile operator’s core network.

For the second case, there is no benefit to the operator (or consumer) to route Internet bound traffic through the UMA tunnel. There may be business or regulatory reasons why the operator wants or has too, but there is no benefit.

In this case, what many operators are doing is offloading non-operator traffic directly to the Internet. I wrote this post some time ago explaining technically what happens. Basically packets destine for the mobile core go into the UMA tunnel, and packets destine for the Internet don’t.

There’s a simple route/decision in the handset to decide which way packets go. This is how BlackBerry phones operate today on the T-Mobile network.

Note that sessions which go over the UMA tunnel can be continued when moving to the 3G network, whereas sessions delivered over Wi-Fi directly (no UMA) will end and must be restarted when moving to the macro network. One of the benefits of UMA is mobility for packet and circuit services.

Anonymous said...

I would also suggest that security & privacy is a major difference between UMA and WiFi access.

Anonymous said...

Thanks for the explanation.

I wonder if you have any comment on the following news published on LTE World Summit website.
Another alternative is the Generic Access Network (GAN). This allows a GSM user and control plane to be transported over WLAN and it has been deployed in many networks today, but overall market penetration is still low. Issues with a GAN based solution again is due to cost as extra GAN controllers (GANC) are required in the radio access network and the mobile has to support extra protocol stacks. The GAN also requires that the IP connection is kept alive - this impacts the mobile battery life and transport of voice and control plane data is non-optimal for LTE.

Steve said...

I had not seen that article. I will have to jump into that discussion too.

As it relates to voice over LTE, the comment about ‘overall market penetration’ being low is irrelevant. Overall market penetration of IMS is low, but people still talk about it as an option for voice over LTE.

The other comments are all one-sided:

- ‘cost of a GAN controller’ – is there an option for LTE voice that doesn’t have any cost? For operators with no IMS infra, they will need to spend for it. Nothing comes for free. I think a better question is what are the operator requirements? Based on those requirements, what is the most cost effective method for delivering telephony? GAN will be at the top of the list.
- ‘the mobile has to support extra protocol stacks’ – every solution for voice over LTE will require new protocol stacks. The question is, again based on operator requirements, what is the lowest risk path to voice over LTE? With GAN, the handset uses the *same* telephony stack used in 2G/3G. GAN is a transport technology, and simply packetizes the telephony messages and sends them across the IP/LTE network. The alternative is to create an *entirely new* telephony stack on the phone (SIP, MMTel?). Now the phone needs to support *two* telephony stacks, the existing 2G/3G stack and the new SIP/MMTel stack. Which is more complicated? Don’t get me started on doing hand-over of an active voice session from SIP/MMTel to 2G/3G…
- ‘GAN requires the IP connection be kept alive’ – What??!?!? The phone will have an active connection to the LTE network with GAN or not. It’s an LTE phone, it must ‘keep alive’ it’s connection to the LTE network. It has nothing to do with GAN. What people may be missing here is this is not ‘UMA’ for dual-mode phones. What is being optioned is using GAN to turn the existing voice network into the core packet voice service for LTE. ‘keeping the IP connection alive’… the phone will constantly be attached to the LTE network for all services. There is no added cost or performance associated with GAN for this at all.

Using GAN to deliver telephony over LTE is very low cost, very low risk and quite pragmatic. If you’re in the business of selling LTE RAN gear, GAN is good. It’s a way to accelerate LTE deployments. If you’re in the business of selling IMS for *telephony*, GAN causes delays (note non-telephony IMS works fine with GAN). As with most things in life, it depends on your perspective

Anonymous said...

Thanks Steve for your patience and detailed explanations.