background image
628 Chapter 16: Voice Techniques
Next, apply modifiers to these figures based on the actual conditions. Types of modifiers to
apply include packet overhead, voice compression, voice activity detection (VAD), and
signaling overhead. Packet overhead can be used as a percent modifier. For example:
20 bytes for IP
8 bytes for User Datagram Protocol (UDP)
12 to 72 bytes for Real-Time Transport Protocol (RTP)
Without using Compressed Real-Time Protocol (CRTP), the amount of overhead is unrealistic.
The actual multiplier is 3. CRTP can reduce the overhead further, generally from 4 to 6 bytes.
Assuming it is 5 bytes, the multiplier changes to 1.25. Assuming that you are running 8 Kb of
compressed voice, you cannot get below 10 Kbps if you allow for overhead.
Voice compression and voice activity detection are also treated as multipliers. For example, use
a 0.125 multiplier for conjugate structure algebraic code excited linear prediction (CS-ACELP
[8 Kbps]). For VAD, use a 0.6 or 0.7 multiplier.
Signaling overhead is an additional consideration. In particular, you need to factor in the Real-
Time Control Protocol (RTCP) and H.225 and H.245 connections.
PSTN-to-IP Traffic Conversion Example
With the information you have collected, you can apply traffic distribution to the trunks to see
how that distribution affects bandwidth. Traffic distribution is based on busy-hour and average-
hour calculations.
Suppose that for the busy hour and the average hour, the distribution of traffic per trunk is 2.64
Erlangs and 2.2 Erlangs, respectively. If one pulse code modulation (PCM) voice channel
requires 64 Kbps, the busy hour will have the following amount of traffic:
2.64 Erlangs
Ч 64 Kbps = 169 Kbps
and traffic during an average hour will be:
2.2 Erlangs
Ч 64 Kbps = 141 Kbps
Therefore, 2.2 Erlangs of traffic carried over IP using voice compression requires the following
bandwidth:
141 Kbps
Ч 0.125 (8 Kb voice) Ч 1.25 (overhead using CRTP) = 22 Kbps
NOTE
The CCDP must account for other modifiers as well, such as call setup and teardown signaling
overhead, Layer 2 overhead, and voice activity detection (if used).
87200333.book Page 628 Wednesday, August 22, 2001 1:41 PM