Monday, October 09, 2006

VoIP Voice Quality - Not There Yet But Not Too Far Away

I have been hearing and reading about lots of complaining concerning the voice quality of a VoIP connection....both residential and business Is this real or memorex (so to speak). Just what are the REAL issues...and where may we (especially businesses) be headed with the maturation of VoIP?

The issue is still that our IP networks are packet based. This is efficient for moving data, but not so good for time and sequence sensitive traffic.

Over private networks, we can adjust the communications elements for session versus packet centric performance. We make sure we have sufficient bandwidth to allow a smooth stream of session traffic (VoIP for example), we adjust prioritization so that session traffic has priority, we change our balancing and routing to insure sessions follow a consistent symmetrical route. The result is a less efficient use of our bandwidth capacity, but a higher quality session for the users. In short, we move away from the purely packet delivery focus and towards a channel like network.

Over the Internet, we lose the ability to optimize our session traffic. The Internet is by design application neutral. The focus is on packet delivery, and each packet is as important as every other packet. At each step in the communications path, the devices are tuned to receive a packet, determine which port to send it out next, and move it on its way. Load balancing across multiple paths, each packet to a given destination may take a unique route. The criteria for success is the delivery of packets error free.

So what is likely to happen? I expect prioritization of session traffic over the carrier networks. They will implement it first for their own services, and the cost of doing so will be recovered from that service revenue. It will be available to individuals, companies, and competing service providers as a premium service. After all, it is a level of delivery above what simple Internet access promises. To work, the carriers will have to agree on respecting each others prioritization when traffic moves from one carrier to another. This will probably be no different than their current method of carrying each others traffic from business sense, and the engineers will quickly work out the technical aspects. The VoIP service companies will scream that this isn't fair, that their service is simply using bandwidth paid for by their customers' access fees. However, session prioritization is not what their customers' have contracted for, so their complaints will be ignored. A new level of access will become common - possibly called Voice Assured or something along that line.

At some point, either a new startup carrier or an existing carrier will decide to market session priority as part of their standard level of service. If sufficient customers switch to get this, the other carriers will follow suit. By that time, most of the networks will have become session prioritized as the standard build.

Big iron will not benefit from all of this.

Some session service providers will lose out to the carriers because their business model relies on the performance of a competitor. Some will step up and pay for session priority so that their customers do not. If they can survive with the reduced margins until session priority becomes the norm, they will retain their customer base.

(Q)Has anyone experienced poor call quality using VoIP?

Yes - of course most people have knowingly or unknowingly. VoIP traffic in all methods of delivery- Skype, Vonage, Cable, IP-PBXs, Peer-2-Peer, softswitches and COs... have varying degrees of voice quality issues in their experiences with VoIP. At least for now.

(Q)Is this a case of poor equipment, poor software, bad connections, or what?

This is a very broad issue. Too many people expect to "just plug it in" and it's going to work- whatever "it" is defined as. The same is true about VoIP due to marketing, misfires, bad judgment, and inexperience.

There are many other reasons too - DSPs which are improving (Fact), software gets fatter which patches the known existing issues and maybe creating a few new unknowns still (My belief), connections- a few in the cables, connectors themselves but everything is relevent to what is defined as VoIP which is just a protocol- what about all those other things to access, control, and transport those packets?

Then - keep in mind that a significant majority of "telephone lines" are copper, TDM based. Longer loops have boosted loop current levels and mixed with IP -- you get echo.

The "list" of issues or causes and effects is just mind boggling. It's not simple or black and white - short answer is "it depends."

Once VoIP can meet those expectations of "just plug it in" then we will in doubt be in a new world of telecom. It's a journey and it will be an adventure for those that tough it out. It will be interesting to see and experience how it all plays out.

(Q)Do you think that less than toll quality voice will be a limit to the growth of VoIP?

No. Less than toll quality isn't a VoIP metric for enterprise or carriers (Big Iron) or the softswitch world either. VoIP as a whole - is improving- at least so says the media. :) Call quality is moving away from what we do for example with software and an appliance to watch voice packets, equipment, and other things... to embedded monitoring call quality within the software itself reporting back across the network. This is significant in when it becomes the norm--- then less expensive solutions to monitor, packet shape, and direct voice packets to their final destinations on time will notably change quality, MOS scores, etc. Who can afford the existing tools other than Big Iron and L-Enterprise ?

(Q)If poor voice quality continues can this cause a backlash against VoIP and a return to Big Iron for some companies?

Not likely. Too much is already invested and proven in the way of the carriers successfully delivery VoIP traffic without the end-user knowing they are in fact on a VoIP segment or call. Ethernet Layer 2 provides a slightly cheaper method of delivery over TDM and since cost is always a factor and as "techniques" improve so will delivery. It isn't likely that a "pull out" will occur.

(Q)What can be done to raise VoIP voice quality to toll grade?

It depends upon which audience is addressed. From where I sit:

1) Training - Certification - Field Experience by those implementing VoIP;

2) Metrics that measure not just MOS but the actual voice packets for jitter, latency, etc and then accountability in the missions of those making the decisions- did we achieve our objective and what were the true costs in doing so and how did it impact us?

3) Timing - hardware isn't getting worse, it's getting better. (That's a fact) The industry is in a learning curve- one that's not going to become stagnant because deliverying VoIP in any form factor is challenging- it's not for those faint of heart. (Implementers)

4) User Expectations - this is one of the greatest failures not just in VoIP but delivery of any telecom / IT service or solution. The expectations are not set and there is a consistent failure of "the meeting of the minds."

5) Benchmarking vs. Hook-Line-Sinker - instead of ramping an entire effort for VoIP cutovers - organizations need to set some bench marking in place first. The temptation to go big vs small because the leaders within the organization need to score a big hit (cost savings).

6) QoS, access, transport, infrastructure - these all apply to any user of VoIP.

Before jumping in, catch up and get into more specifics about what is wanted, methods of delivery, metrics, how VoIP is planned before inking a deal or making a change with someone promising them something better, faster, cheaper - as with anything.

VoIP is here to stay. Sink, swim, or paddle - it's rewarding, it's a beast to manage, challenging and very rewarding for those who are prepared.


Post a Comment

Links to this post:

Create a Link

<< Home