Article on VoIP Quality Factors
I want to point out a really good column online as part of the ongoing Convergence Newsletter published by Network World, written by Steve Taylor and Larry Hettivk. Steve's a longtime friend and colleague, and while I've never met Larry, these two always do a great job of covering important trends and issues around converging technologies.
Factors that affect VoIP call quality
What can make VoIP call quality unacceptable?
Convergence Newsletter By Steve Taylor and Larry Hettick, Network World, 08/28/06
Following up on the VoIP quality issues that we discussed in last week's newsletter, this week we'd like to take a deeper dive into some of the possible problems that can make VoIP call quality unacceptable.
I'll touch on the highlights to give you a taste, but you reallty do need to go read this well-written column.
CODECs are important. G.711 and G.729 are the two most commonly used in VoIP. While they don't mention it, I'd suggest that the Global IP Sound (GIPS) CODEC probably beats both in quality, although I haven't seen any specific MOS comparison testing. GIPS is coming on strong as a leading CODEC choice.
Packet loss or discard is a quality concern. Since the CODEC does compression, anything that gets discarded becomes a problem. The more loss, the lower the quality of the call. Remember, IP is a best efforts protocol with no guarantees.
Latency (or delay) is another quality issue. Latency is directly tied to the path through the network.
Jitter is variable latency. It's almost a non-issue in bursty data traffic because the user never experiences it. In a real-time two-way conversation, it makes for herky-jerky sounding voice calls.
Technorati Tags: VoIP, Convergence, Steve Taylor, Larry Hettick, VoIP quality
Email This!
Digg it!
Sphere it!
Del.icio.us
Reddit!
Newsvine
Comments
Good point Ken. GIPS' codecs and soundware were built for packet-based communication versus the circuit-based G711 and G729. GIPS' iLBC is even an IETF (internet) standard, as opposed to an ITU (telecom) standard (such as G711 & G729). The GIPS codecs are better on IP networks, especially when packet loss comes in to play, i.e. almost always. See GIPS comparisons in MOS scores on their website:
www.globalipsound.com
Posted by: Rick | August 29, 2006 2:45 PM
Great point that I failed to mention, Rick. GIPS was designed for packet, rather than circuit switched networks. That's certainly a technical factor to consider. And thanks for letting me know they have some MOS comparisons online. I think that warrants some review.
Posted by: Ken Camp | August 30, 2006 9:01 AM
You may know this, but many people don't. The MOS scores between narrowband and wideband codecs are graded on DIFFERENT scales. In other words, a 4.2 for iLBC is not the same as a 4.2 for iPCM or iSAC.
Posted by: Rick | August 30, 2006 2:32 PM
GIPS is becomming more and more a leading player in the field. There TECHNOLOGY is state of the art and from a STRATEGIC point of view they are well positioned (f.ex deal with skype opens hardware AND wifi market on a global scale). Open question for me is still the FINANCIAL part. The figures until know look excellent but it's unclear how they can leverage from deals like skype. Do they have a recurrent financial model f.ex. Maybe this information is in the market but it was never clear communicated then. Anyway, if they succeed in having a leveraged recurrent model then we are having a real winner here with future financial posibilities to defend succesfull their current technological leadership.
Posted by: Jan | September 1, 2006 3:22 AM
Jan, Thanks for the thoughtful comments. I too wonder about the financials. Rick, who also commented here, had asked me to talk to Jan, their VP at the recent VoIP Developer Conference. Turned out Jan came in nriefly to speak, but due to a brand new baby, couldn't stick around and talk. I did chat with a couple of their tech folks, but I'm hoping to get some quality conversation time with them at the upcoming IP Telephony Expo in San Diego. I'd like to understand their whole model and approach better. I see them assuming a more dominant position all the time.
Posted by: Ken Camp | September 4, 2006 9:31 PM