Slashdot Mirror
Cable VoIP Sounds Better Than Some Landlines
A. G. Bell writes "A recent study that looked at the quality of phone calls came up with some surprising results. Ars Technica reports that while 'traditional' VoIP call quality lagged behind landlines, service from cable ISPs was much better because of their use of PacketCable: 'VoIP from the cable companies actually surpassed the traditional phone network in reliability, meaning that the service was more often available and connected calls without dropping them. Cable providers also led the way in audio quality; the top firm in Keynote's study actually turned in an MOS of 4.24, above most real phone networks.'"
In my little hometown they still use the original ancient phone lines that leak signal like crazy. In fact, you could easily hear other conversations if you paused while talking on the phone. I'd guess that a majority of towns are using lines that are at least 30-50 years old still.
This seems to me like the ISP gets an advantage because of this PacketCable thing -- something I'm sure they will not be licensing to their VoIP competitors like Vonage. Not surprisingly, these 'other' VoIP providers fared worse then the ISP-provided VoIP service.
That's the point. When a VOIP call is made from a cable ISP, the call stays on the cable ISP's backbone(but not the regular net. A wicked huge Intranet would be a better analogy) for as long as it can. Some cable providers created additional plant lines just for this. With Vonage and friends, they hop on the normal internet and do the 20 + jumps of fun. Think of PacketCable as an express lane for the Cable ISP's calls.
"Useless organic meatbag" -HK-47
I should probably give the obligatory "I work for Comcast, but just as a dispatcher" warning, so I know a bit more about this than most people. :-)
Yes, but until any one cable company has coverage to every home in America, a call from NY to CA will most likly traverse another provider's network
Right, unless both sides have (for example) Comcast's VOIP, there will be a hand off between providers. But all phone companies pay when a call transfers to another provider's systems. The amount per call is next to nothing, but considering the number of phone calls made at any one time, it adds up to enough that I know Comcast has laid cable through areas they don't service just to carry their own VOIP calls. Same for cell phones(ever wonder why your phone always homes in on your provider's towers even if another one is closer?). So if Comcast hands off the last bit of a call to say AT&T's network, Comcast would pay AT&T as they would if the call went to Comcast's network.
As for the lawsuit, isn't that the whole debate about the net neutrality issue? What is different from SBC trying to extort more money from Google for data passing over its lines than AT&T trying to extort more money from Comcast or RoadRunner for the same reason?
Different issue. Over Comcast's packetcable thing, the data of the call only goes over Comcast's backbone(and no other part of the net), then it is handed off as a normal phone call.
"Useless organic meatbag" -HK-47
Cable providers also led the way in audio quality; the top firm in Keynote's study actually turned in an MOS of 4.24, above most real phone networks.
If I recall the DOCSYS standard correctly (that's the one for cable settop boxes), the framing provides the phone company TDM-style 8 kHz synchronized clock, and the phone signals are carried as full-rate uncompressed bytes.
In other words, POTS-over-cable is a 64 kbps synchronized digital signal, identical to what's carried on the phone companies' own ISDN, T, and SONET carriers, and is switched onmodified on and off the rest of the digital network unmodified. The A-D conversion happens in the settop box. It's like having your POTS phone at the switching center within wire-feet of the multiplexer. (The clocking is also good enough to encode analog signals from FAX and 56K computer modems. It has to be, as a side-effect of the need to time the upstream packets properly.)
POTS, on the other hand, is A-D converted at a central office or a "remote terminal" (in a box at the curb) and carried the rest of the way - blocks to miles - in one of a bundle of wires. This is subject to crosstalk, distortion (selective delay and attenuation of higher frequencies), and a number of other pathologies that lower the signal quality.
So it is not at ALL surprising that cable POTS signal quality beats telco POTS. Cable's signal is about as pristine as you can get.
(And VoIP isn't in the same ballpark, due to both compression and timing uncertainties.)
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
You can absolutely test 911 from any phone line. Simply call 911 and immediately mention to the operator you are making a test call to verify address information. Tell her what you think your address should be in their system and he/she can confirm this for you. As long as you make clear right off the bat you are on a test call, there is not a problem.
John Susek
VoIP uses whatever codec it's configured to use, which could be great-sounding G.711 alaw/ulaw (as used on the real phone system) or GSM (not so good).
The problem isn't with bandwidth. Anytime you place a call on a land line (non-local), your signal on the analog POTS line is digited (8KHz, ADPCM, uLaw encoded) at the FXO (foreign exchange office) and transfered over a channel on a T1. 8KHz is plenty of bandwidth to transfer the human voice (Bell labs figured this out in the 50's)... With VOIP, extra bandwidth on one end doesn't give you a thing if you're calling a non-VOIP phone on the other end... eventually the signal gets transfered over a T1, at 8KHz (the lowest common denominator). The real problem is that VOIP is packet switched, without a guarantee of delivery... and as you know with IP, you can also run into issues of packets being received out of order. The real value in VOIP isn't QoS (it won't be for years to come)... the real value is in being able to deploy voice services over an existing IP infrastructure... no more clunky and expensive PBX's!
Now, if you're placing a VOIP to VOIP call (I assume this is what the test was... didn't read TFA), there is no 8KHz constraint and throughtput can theoretically be higher.
# man tar
Before I comment, I'd better post my credentials to say what I'm about to say: I am a co-author on many of the PacketCable specs, wrote the standard text on the subject, and also run a little company that sells PacketCable security software.
So, having said that, I would like to point out that your comment is accurate but may be a little misleading. It may indeed be true that today, and for most cable ISPs, the call stays on their network. But PacketCable was not specifically designed to be that narrow. Its architecture allows lots of things that cable companies have so far mostly not chosen to do. But, for example, there is no reason at all why the service has to be provide by a cable company. Sure, the cable company controls the pipe into the house (and the quality of service on that pipe), but there is nothing at all to prevent an ISP that decides that it doesn't want to get into the telephony business (and telephony could not remotely be described as easy) from contracting with a "real" telephony company so that that company provides service, with all the usual quality of service controls, over the ISP's network.
To give a completely and obviously hypothetical example. Instead of deploying telephony itself, Comcast could have chosen to have Qwest run a Comcast-branded VoIP service over Comcast's network, including the last-mile access network. That service could be given exactly the same quality-of-service guarantees that Comcast has chose to give to itself, and presumably both Qwest and Comcast would receive a cut of each phone call.
The corollary is that third-party providers (the Vonages of this world) do not have agreements and service-level contracts with Comcast. This means that their calls travel over the Comcast network using "best-effort" instead of some kind of guaranteed quality-of-service labelling. In particular, between the subscriber and the cable comapny, Vonage-like services travel over the ordinary standard primary DOCSIS flow from the cable modem, sharing it with all other traffic from that modem; PacketCable calls use special flows that have guaranteed latency-and-jitter limits specially designed for voice. Only the cable company can create and use those flows. (For the gory technical details, look at the DQoS [Dynamic Quality of Service] spec available at www.packetcable.com.)
The first time I connected on Skype (from Shaw in Manitoba to Telus in Alberta), the voice quality was amazing. My father instantly noted that I sounded like I was "right there". Just like watching HD for the first time, other phones (especially mobiles) sound so much worse since we've started runing Skype.
And now that I took time to read the Wiki link, I see that is exactly what it says. So for commercial VoIP, you would specifically NOT use hot-potato routing.
"Hot lesbian witches! It's fucking genius!"