[3GScottishUser]

From PC Advisor (20/03/2006):

Mobile VoIP needs high-speed uplink
On its way, but not any time soon


Internet telephony over mobile phones is on the way, but don't expect many commercial offerings until operators have made a key network technology enhancement, probably toward the end of 2007 or later in most parts of the world, according to a senior executive at Lucent.

The problem with providing VoIP (voice over IP) service over mobile handsets today is the uplink, which is too slow to support high-quality voice calls, according to Lucent chief marketing officer John Giere.

To increase uplink speeds, operators of GSM (Global System for Mobile Communications) networks, which dominate Europe and many parts of Asia and Latin America, will need to upgrade their networks with HSUPA (High Speed Uplink Packet Access) technology, he said.

"HSUPA will give operators the bidirectional capability they need to run real VoIP," Giere said in an interview last week at the CeBit trade show in Germany.

However, the Lucent executive doesn't expect the high-speed technology, which is currently being standardised, to become commercially available until the latter part of 2007 or early 2008. Operators are presently busy rolling out the downlink counterpart HSDPA (High Speed Downlink Packet Access).

How that timing fits into the VoIP over mobile plans of Skype and Hutchison 3G is unclear.

At the 3GSM World Congress in Barcelona last month, the two companies announced a partnership to provide the world's first commercial VoIP service for mobile phones. The companies aim to begin offering service in select markets as early as this year, said Christian Salbaing, managing director of European telecommunications at Hutchison 3G, in an earlier interview.

Though Hutchison operates an IP-based network, it has not rolled out HUSPA.

In addition to HSUPA, operators planning mobile VoIP services will need to 'flatten' their networks by reducing the number of components and using IP wherever possible, according to Giere. "The number of network components you have also contributes to network latency, which is a big issue with VoIP," he said.

Most GSM operators today have legacy circuit-switched networks, which tend to slow the flow of IP traffic because of their numerous network components and conversion processes, according to Giere.

In response to the need to reduce network components, Lucent has introduced a new base station system that collapses a series of network architecture layers into one component.

Lucent is also collaborating with Samsung in a project aimed at developing SIP (Session Initiated Protocol) client software, which is essential for offering VoIP for mobile handsets, Giere said.

Until now, mobile operators have largely dodged the great VoIP debate, trying to squeeze every possible cent out of their largely amortised circuit-switched networks before investing in yet another new technology.

But should they be interested in VoIP? "Absolutely," said Giere. "Efficiency is one reason; operators can significantly increase their bandwidth utilisation with VoIP. Applications are another; an IP environment is all about creating a rich set of applications."

Another reason, especially for those operators that are net payers of international roaming services, is the ability to use VoIP to undercut high intra-carrier network usage fees.


http://www.pcadvisor.co.uk/news/index.cfm?newsid=5851

[Ben]

Yeah, fixing the upstream is an obstical. VoIP over the current 3G networks will have latency issues, though HSDPA will help on the downstream side. It'll work, but we'll need HSUPA before the quality rivals GSM me thinks.

[3GScottishUser]

So it'll be like talking in a tunnel...... somewhat similar to normal 3 voice calls as I recall, those exhibited lots of latency.

Why on earth are they going to launch a service that is not yet perfected?

Have they learned nothing from the last 3 years?

[Ben]

I'm pretty sure that in vast swathes of 3's coverage there are no latency problems on voice calls anymore.

I guess no innovative step is ever perfect. Hell, most of the current mainstream offerings aren't perfect - like music downloads! By floating the balloon early they'll be able to gather feedback and make improvements before things go mainstream. It certainly wont be mass market from the off, and Three and Skype know that.

[Hands0n]

I've not heard the "3 tunnel" since around six months after I bought my e608 pair. Voice quality has been as good as any of the 2G/2.5G counterparts that I carried in tandem with my 3 handset.

I do not believe that there is anything "imperfect" about VoIP, it has been around in one form or another for a very long time now. My earliest touch of the technology was with Cisco product going back at least five years now. In one instance it was operating over a general-purpose 64Kbps intra-European IP-based WAN carrying all sorts of other corporate data - no CoS just some simple router prioritisation, just dropping the VoIP into the serial stream. It worked as well as the older analogue voice channels that it replaced! All very pioneering stuff, and a good example of what can be done, simply(!). Scaling the technology for a national mobile operator given todays state of the art should not be a major challenge.

Latency is going to be caused by more than merely the uplink speed -and if it is 64kbps I doubt that it will be the major limiter - unless the VoIP is configured/designed to need more than that, and why would it be? Is Lucent seriously suggesting that we need Megabits of uplink speed merely to get VoIP working properly? I don't buy that at all. But then surprise surprise Lucent are manufacturing the "solution" to the "problem".

Oh go on then, call me cynical :D

[Ben]

Latency is going to be caused by more than merely the uplink speed -and if it is 64kbps I doubt that it will be the major limiter - unless the VoIP is configured/designed to need more than that, and why would it be? Is Lucent seriously suggesting that we need Megabits of uplink speed merely to get VoIP working properly? I don't buy that at all. But then surprise surprise Lucent are manufacturing the "solution" to the "problem".
I think you might have confused latency and bandwidth slightly.

HSDPA and HSUPA make big increases in bandwidth, yes, but VoIP doesn't need the extra bandwidth that HSPA technologies bring. However, HSPA technologies are claimed to dramatically improve the latency problems currently experienced on WCDMA networks by making their data capabilities more efficient.

To the best of my knowledge the latency on GSM GPRS networks is lower than the latency on WCDMA networks. Lower being a good thing. The only reason GSM GPRS seems slower than WCDMA is due to the lower bandwidth, meaning we have to wait for longer for larger amounts of data to be transferred to us. You may have noticed that browsing text WAP pages on WCDMA is actually slower than GSM GPRS at the moment: that's because the time to contact the server and download the small amount of data is less. HSPA might at least level up the playing field.

[Hands0n]

Bandwidth can impact latency, particularly if the availability of the bandwidth is low (i.e. congestion causing queueing or discards) and there is no QoS or other prioritisation of the VoIP data. I recall seeing these kinds of problems with the WAN link we used for VoIP, it was general purpose and until we got the prioritisations right the quality could be affected by the other data. On other WANs with higher bandwidth/availability VoIP would more readily work without prioritisation (although we did use it to assure the VoIP data).

But in a mobile handset I would not expect such a congested condition to exist unless the user is at the same time as using VoIP conducting some huge upload to the network.

It is my understanding that GSM GPRS networks have a high latency, higher than that on WCDMA as it currently exists, although the latter is still quite poor in such terms. Whenver I've used teletype-style apps over GSM GPRS the lag is quite noticeable. But I've not tried the same tty over WCDMA yet so cannot compare the experience. It is good, however, that HSPA technologies seek to improve upon not only on bandwidth but reduce the latencies significantly. Of course, that has to be a good thing in anyone's book.