Suppressing silence saves money

If you are calculating the economics of running voice over frame relay, don't forget to figure in the possible savings of using voice activity detection. VAD - which we once called "silence suppression" - can reduce the amount of bandwidth you need for voice conversations by roughly 50%.

VAD reduces your bandwidth requirements by not generating traffic during periods of silence in an active voice connection. The 50% savings comes from the simple fact that in most conversations, only one person talks at a time, leaving the other half of the voice circuit idle. Also, those pregnant pauses in the conversation can be removed. Note that while VAD frees up network bandwidth when one person isn't talking, it is still possible for two people to talk simultaneously and to do so without blanking each other out, as happens when using some speakerphones and cell phones.

VAD could be particularly economical if you are on a usage-based pricing plan: You certainly don't want to pay for transmitting silence. But even if you're using a fixed committed information rate (CIR) plan, activating this feature in your voice frame relay access devices or in routers with voice support allows you to fit more conversations within the CIR.

Note that the typical oversubscription concepts apply to frame-based voice with VAD. The larger the number of conversations you have sharing a single virtual circuit, the greater the odds are that the traffic in each direction will average out to about 50%. So, as you add more conversations to a circuit, you can add proportionally smaller amounts of CIR to accommodate them, because the bandwidth savings are also growing.

VAD isn't a new idea. Digital speech interpolation has been around for years, and time assigned speech interpolation preceded DSI. But these mechanisms suffered from a drawback called clipping. Clipping is that annoying phenomenon that cuts off a bit of speech in the instant it takes for the transmitter to detect that there is speech present. It is nearly impossible to eliminate clipping in a traditional circuit-switched voice conversation. Using circuit switching, you don't turn on the transmitter until sound is detected, and by then, a piece of the speech has been clipped off. The only way to totally eliminate clipping would be to detect the sound, then move backwards in time by a few milliseconds to turn on the transmitter just prior to when the person started talking.

Going back in time, you may astutely note, can be a problem! But this is where Tachyon Transmission Mode, a.k.a. TTM, invented by Distributed Networking comes into play. Intrigued? Visit www.webtorials.com/ttm to find out more about TTM. Or you can tune into our next newsletter to find out how packet voice technology eliminates clipping. Or do both!

---

Steven Taylor, consultant and broadband packet evangelist, and Joanie Wexler, an independent networking technology editor and writer, team up to bring you this analysis and commentary. Taylor specializes in education and market analysis, and Wexler adds incisive reporting and research. For more detailed information on most of the topics discussed in this newsletter, connect to www.webtorials.com, the first Web site dedicated exclusively to market studies and technology tutorials in the Broadband Packet areas of Frame Relay, ATM, and IP. Feedback and additional topic ideas are welcome. Please contact taylor@webtorials.com or joanie_wexler@mindspring.com.

Bringing Fibre Channel to the mainframe
Network World, 01/18/99

Talk is cheap with frame relay
Network World, 01/18/99

FRAD buyer's guide and review
Network World, 01/18/99

Frame Relay Net Resource: primers and more
Network World Fusion