Mobile VoIP is probably going to be trumped down by IMS. It’s only a matter of time. When carriers will see VoIP as a real opportunity, they will embrace it, and that embrace will be in the form of IMS.
While that is true, I believe we’re still a few years before mobile VoIP will be interesting enough for carriers. But don’t take my word on it – ask a better: Martin Sauter.
In a recent post of his, Martin compared VoIP with GSM from the point of view of bandwidth and network utilization.
I’ll try to sum it up in the table below, although I strongly recommend to read his whole post, as it contains a lot of additional useful information:
| Connection type | Network utilization |
|---|---|
| Circuit Switched | 28 kbit/s |
| Packet Switched | 64 kbit/s |
How did Martin get to these utilization values?
- An AMR voice codec takes 12 kbit/s for a full rate voice channel
- Multiply it by 2 (as there is both incoming and outgoing voice)
- Add the overhead of the payload headers (none for circuit switched, IP headers for packet switched)
- Add the overhead for error correction and detection (which can vary depending on network conditions. Here I think Martin added the worst case scenario)
[Note for advanced readers: Yes, you can try and use IP header compression in packet switched connection. But it is not really used today on HSPA or UMTS networks. In LTE I'm sure it is going to be different.]
When you look at the utilization table, it is no wonder then that mobile operators don’t want people to use mobile VoIP: It eats up a lot more of their network than the same circuit switched call, and at the same time reduces their revenue.
Another point to take into consideration is that while it seems that everyone is going VoIP, in practice there’s hardly anything ready at the present time. Take Google, for example, who is doing a lot of noise as they are “embracing VoIP”. Andy Abramson noted in his blog recently that while this is true in general, there’s nothing ready yet:
A quick read of the official Google blog says so. But other than SiPDroid and a Gizmo client attempt, nothing really is there yet that does the trick over WiFi.
Google Android may be ready, and it is embracing VoIP, but the applications supporting VoIP are still not ready.
Another important aspect is the quality of experience. John Gallagher, who can be considered as an expert as he comes from long-time voice codec vendor GIPS, tried using Mobile VoIP as much as possible in his last trip to Europe, and found the experience somewhat “lacking”:
… The quality of the experience is important and while I’m a believer in VoIP it’s easy to see why VoIP can, and has in some cases, got a bad name. My experience with some applications is less than good (none of them GIPS’ customers I might add.) When a free call is placed, people may a tolerance for inferior quality – that is until a better application or client comes along. Bad voice quality drives away customers and it’s a great way to push new customers over to your competitor.
Bottom line: Yes, you can do Mobile VoIP already. And yes, people are already using it to reduce costs. But it’s still not prime time for Mobile VoIP. Not this year and not next year, at the very least.
And when it does happen, it might just be IMS, and not Mobile VoIP, that rules that world.
Tags: AMR, Android, Andy Abramson, bandwidth, Circuit Switched, GIPS, Google, GSM, HSPA, IMS, John Gallagher, LTE, Martin Sauter, Media, mobile operators, mobile VoIP, Packet Switched, quality, quality of experience, SIP, UMTS, wifi
Comments and trackbacks
1. Dave Michels | September 29th, 2009 at 11:12 pm
One factor this doesn’t show is periods of silence or one way audio.
Packet VoIP does not transmit silence, so over a longer conversation the average bandwidth will drop.
Not sure how the wireless carriers deal with full duplex today, but if they are honoring full duplex (and half the conversation is always silent), then the numbers may even favor VoIP
2. Tsahi Levent-Levi | September 30th, 2009 at 9:26 am
Dave,
There’s one aspect that I think might be of an issue here – while there are silence periods in calls, can an operator design his capacity around that assumption or should he have enough bandwidth on a given cell to fit X calls without any silence?
I think this will break both the quality AND reliability assumptions set out by Garrett…
And it still means that VoIP takes more bandwidth than operators may be willing to sacrifice for “mere” voice calls at the moment.
Tsahi
3. Mike | October 9th, 2009 at 2:15 am
Interesting,
As one who has evaluated several VOIP over mobile applications, each was less than satisfactory given lack of quality of calls, similar to how first VOIP systems provided less than crystal clear calls.
VOIP over mobile applications are an issue as I learned, just check the TOS agreements with TMobile and AT&T and you will see they do not allow voice over their optional data networks. I doubt the carriers will allow anyone other than themselves to provide VOIP, and since it will reduce their profits, they will not be in a hurry to adopt VOIP while COIP is answer.
Trackback this post | Subscribe to the comments via RSS Feed