VoIP Cell Phones Coming

bp33 writes "Wireless Newsfactor is running a story about how the wireless vendors are climbing over themselves to get Voice-Over-IP cell phones. You might ask "why bother? We already have wireless voice now." But with an open platform for wireless (Symbian, JavaPhone etc), your "voice" (er .. audio) just becomes bits that your programs can manipulate before sending."

Comments

and...?

[Scaebor]

You might ask "why bother? We already have wireless voice now." But with an open platform for wireless (Symbian, JavaPhone etc), your "voice" (er .. audio) just becomes bits that your programs can manipulate before sending."

and... this is the explanation? What am i missing here? This is a serious question. With all the ideas that you people come up with there must surely be some good reason for having a phone that has ip integrated into it.

I'd rather

[URoRRuRRR]

I'd rather have phones that worked really well first. I'm tried of having half of my calls dropped.

Latency & Jitter

[chill]

Voice uses circuits for a reason -- latency and jitter *must* be controlled or the conversation goes to hell.

There has to be more to wireless VoIP than simply 3G+ data -- it must be able to control the timing of the arrival of packets.

No, you can't buffer it. Voice conversations are realtime interactive. Fat packet sizes don't help, either. There is a limit to how long you can spend processing the data into and out of a packet before you screw up the timing.

They have a LONG way to go before this will be realistic.

Re:Latency & Jitter

[WolfWithoutAClause]

Voice uses circuits for a reason -- latency and jitter *must* be controlled or the conversation goes to hell.

True, however if the network is lightly loaded, IP introduces negligable latency and jitter. VOIP is already being used for long-distance telephone calls.

There is a limit to how long you can spend processing the data into and out of a packet before you screw up the timing.

True. However we are not talking about big packets. Normal telephone quality is only 64kb/s (56kb/s in the US). The reason they are going with VOIP is compression- you can compress the date down by a factor of perhaps 4 fairly easily; partly because you can compress the sound by that much quite quickly, but also because on most telephone conversations only one person talks at once. That's important in a wireless phone network.

Re:Latency & Jitter

[chill]

True, however if the network is lightly loaded, IP introduces negligable latency and jitter. VOIP is already being used for long-distance telephone calls.

True, but no phone network is lightly loaded, except at 2:00 a.m. or so. The phone companies have oversubscription down to a fine art.

The VoIP compression kicks in on the backbone transit, as opposed to the actual endpoint allowing more data to be multiplexed in. Yes, silence supression, comfort noise and single/full-duplex play a big part.

Actually, many of the phone networks are doing the VoIP modulation at either end, then using ATM switches in between. So, in essence, there is still a circuit. However, MPLS is starting to be a big buzzword and companies are starting to deploy it as opposed to Frame Relay and ATM. Still, ATM is king when it comes to these types of applications. Real QoS, real ToS, can't be beat.

My Programs?

[evilhayama]

Just consider the possibilites of what I could do to enhance my telephone calls now... Get some mp3s in the background putting my phone call to a soundtrack, maybe adding sound effects here and there to spice up the conversation? ^_^ (not to mention the aforementioned voice morphing)

More seriously, does this mean i could encrypt my phone conversations with fellow terr... associates?

Re:Why IP?

[alienmole]

I sympathize with your perspective, but I don't think it's as simple as you suggest. There are a whole host of factors that converge to make IP attractive to a wide audience.

One important factor is that many people saying "use IP" are really saying "use a standard, packet-switched protocol". There are some good reasons for doing this, too. But it just so happens that in today's world, only one protocol fits that bill. To demonstrate the impact of this issue, from your comment, I can't tell whether you think packet-switching in general is being overused, or just IP. The two have become virtually synonymous.

Technically, it's not that difficult to support replaceable network and transport layer protocols. Novell did it, a long time ago. That would allow more diversity in the choice of protocol. But the problem is standardization - everyone would have to agree on the infrastructure to support that. To summarize wildly, it was easier just to agree on a single network and transport layer protocol suite, than on a standard for making those protocols pluggable.

Having agreed on the standard, we're now stuck with a situation in which it tends to make sense to do anything that needs packet-switching, using IP. Even if it doesn't seem to always make technical sense, it often makes economic sense, because you can reuse existing technology, expertise, and infrastructure. Networks effects can be wonderful, but they're also tyrannical.

News flash: VoIP actually works.

[XNormal]

You must be confusing "voice over IP" with "crappy free telephone calls over the public internet". Voice over IP in a controlled private network can have strict QoS guarantees on latency, jitter and packet loss. VoIP is actually used by many millions of people, most of them don't even know it.

Cellular networks use voice compression codecs that must accumulate a complete block of samples before compressing and transmitting it. They also use heavy error correction. Both of these factors introduces a very significant latency. If the voice compression blocks, error correction blocks and VoIP packets are all in sync some of these latencies overlap instead of adding up and it may not add any significant additional latency.

Re:The obvious answer: convergence

[mesocyclone]

If convergence is the only reason, why not just do it all over TCP/IP or even HTML? (Answer: engineering reasons - same argument one might make against IP itself!).

Furthermore, IP is a low level protocol. It doesn't guarantee interopterability! To have interoperability, one needs all levels of the protocol stack to be compatable, and the hardest one there is the applications level, not the various transport levels. This means, for example, that if your phone does messaging, that it interoperate with other phones and/or hosts that provide messaging service. IP is the least of your problems in that regard!

I could see having, IPv6 addressability for all phones, but that is not the same thing as actually using *IP* as the transport mechanism.