Re: Dr. Ari Jaaksi on Maemo 5
 

hmm, that contradicts what i read while the local telcos where building umts and had issue with coverage indoors in the cities.

but then funny little details about these things keep sneaking up on me and altering my understanding of things...

Re: Dr. Ari Jaaksi on Maemo 5
 

It does not necessarily contradicts your information. Telcos do have issues with indoor coverage and umts. They have similar issues with GSM 1800, a bit different issues with gsm 900.

Re: Dr. Ari Jaaksi on Maemo 5
 

ok, the reasons for that problem is just more complicated then reported in other words.

Re: Dr. Ari Jaaksi on Maemo 5
 

Yes, you made some good points with your compares of SS7 & SIP over HS*PA however despite that the codecs wouldn't saturate the connection. Most data plans provide far more than 64 kbit/sec upload.

Now lets say 64 kbit up, and 64 kbit down, that is approx 1 MB per minute. If you have a data plan of 1 GB/month for 30 EUR, you pay 0,03 EUR for 1 MB. For international data calls you can't beat that.

That also depends on whether you use TCP or UDP.

I'm not sure they're blocking SIP.

But they're not providing unlimited voice calls either. And you know why they don't do that. There is no incentive.

In a healthy environment, telcos would compete with each other but we don't see this in the GSM/3G market very much.

Yes, there is. Nobody knows if the contract holds up in court. It is gonna be a lot tougher though.

And minors I recommend to sign any contract they're getting. Because these aren't legally binding because they're under age! :D

One thing is sure: for international data calls, SIP is worth it.

Re: Dr. Ari Jaaksi on Maemo 5
 

Same with WiMAX.

In practice it differs a lot where and how big these coverage issues are. For example, in the Netherlands KPN and Vodafone provide very good coverage while T-Mobile (ironically also the one who got the iPhone 3G contract) has its fair share of problems.

BTW, it isn't always the telcos fault if there is bad coverage in a specific building. You will find that often there is a lot of interference from materials such as metals. Or its something related to some government law...

Re: Dr. Ari Jaaksi on Maemo 5
 

The problem tends to be one of latency, rather than bandwidth. I've used SIP with the G729 and GSM codecs quite happily (as an experiment) over GPRS. But the same wasn't practical over 3G (UMTS) because of the latency. However, HSPA has much lower latency than UMTS, so should be more feasible.

SIP is a protocol, not a voice technology. And saying that it uses more bandwidth than voice is a misnomer. Voice on POTS is transmited at 64k, either u-law or a-law, and this is _usually_ the default codec used for SIP-based VOIP providers.

Voice over GSM uses, unsurprisingly, the GSM codec, which was created for transmitting voice over the limited bandwidth. So comparing u-law over 3G to GSM isn't a fair comparison.

Considering the minor difference in compression, the G729 codec offers a surprising quality boost over GSM. Plus, it is supported by the N8x0 SIP client and GSM isn't.

Re: Dr. Ari Jaaksi on Maemo 5
 

I stated this a bit earlier in the thread, but in addition to what you've already stated, seems like a few of us are seeing the picture beyond the IT.

Very cool...

Re: Dr. Ari Jaaksi on Maemo 5
 

3G is not UMTS. UMTS is considered part of 3G, like HS*PA.

Didn't know this specific difference. This is very interesting, and good news.

I suppose Jerome was referring to the most popular audio codec for SIP.

Re: Dr. Ari Jaaksi on Maemo 5
 

You don't understand. If the cell is only serving you, your figures are correct. In real usage, and more so if everybody tries to use voip, umts (or hsdpa, which really is umts with a better modem) degrades. You won't get that upload.

What is so difficult to understand? Radio is simple: you have one medium and it has a maximum capacity. Whatever tricks you use, when the capacity is exhausted, that's it.


I am not comparing codecs, I am explaining the fundamental problem telcos have with sip over air. It is not a technological problem, it is a marketing problem. It is a bit like the famous "tragedy of the commons": when customers pay 10 times less per calls, some save 90% (which is not really a problem) and some think they can phone for 10 times as long. The system is not financially tenable in the second case, especially if these users chose wasteful codecs (out of ignorance usually).

You don't have that problem with wires, because it is not a shared medium. When some users bog down their connection running p2p clients like mad or having skype relaying the traffic of half the planet, it is not a problem for the rest of the users. You have a problem with data over air, because when one user abuses the cell, everybody's connection goes dripping. And building additional cells costs money.

It works when the cell is not overloaded. When the cell gets busy, latency degrades. I have used sip and skype over umts without much problem. With hsdpa, I can even use google video talk.

I don't think that latency would be much different between umts and hsdpa. It's the same system, really. hsdpa just uses a more efficient modulation.

Re: Dr. Ari Jaaksi on Maemo 5
 

That's right. A problem we have doing the RNP (Radio Network Planning) for 3G networks, is that the cell capacity per user decreases with the number of users (they share a fixed bandwidth), but also the cell size become smaller as the number of users in the cell grows. In some cases we saw some zones lossing the signal (becoming black zones) when the number of users in a cell increases significantly.

This is why, in Europe when they have a special event like a football cup final, the operators install in the surroundings of the stadium temporary cells to cope with the increase traffic and the high number of users per cell.