IP Trunk3.01 (and later) DSP profile settings 163
2. Voiceplayout maximum delay.
3. Fax playoutnominal delay. This can range from 0 to 300 ms, with 100
ms as the default size.
As discussed in "Adjust jitter buffersize" (page 136), lowering the jitter buffer
size decreases the one-way delay of voice packets;however, setting the
jitter buffer size too small causes unnecessary packet discard.
If it is necessary to discard to downsize the jitter buffer, first checkthe
delay variation statistics. Obtain the one-way delay distributions originating
from all source IP Trunk 3.01 (and later) sites, using the measurements
outlined in "Measure intranet QoS" (page 139) or "Post-installation network
measurements" (page 164). Compute the standard deviation of one-way
delay for everyflow. Some traffic sources with few hop counts yield small
delay variations, butit is the flows that produce great delay variations that
should be used to determine if it is acceptable to resize the jitter buffer.
Compute the standard deviation (s) of one-way delayfor that flow. It is
recommended that the jitter buffer size should not be set smaller than 2s.
Silence Suppression parameters (Voice Activity Detection)Silence Suppression, also known as Voice Activity Detection (VAD),is
enabled by default on a new IP Trunk3.01 (and later) node. Enable/disable
VADusing the Enable voice activity detection checkbox on the TM 3.1 ITG
Node Properties -- DSP Profile codec Options tab. See Figure 55 "DSP
Profile Codec Options tab" (page 246). To change the current DSP VAD
state to match the current VAD configuration, re-transmit card properties
from TM 3.1.
When silence is detected, the IP Trunk3.01 (and later) node sends a flag to
the destination IP Trunk3.01 (and later) node that denotes start of silence.
No voice packetsare sent until the silence period is broken. There are two
parameters that control Silence Suppression, as follows:
1. Idle noise level. This is set at a default levelof –65 dBm0.
2. Voiceactivity detection threshold. This is set at a default of 0dB. Voice
packets are formedwhen the audio level exceedsthe idle noise level by
this threshold value.
These default parameters are suitable for most office environments.
Increasing either of these two parameters lowers the amount of IP traffic
generated, but increases clipping and dropped packets.
Disable Silence Suppression at tandem nodesSilence Suppression introduces a different concept of half-duplex or
full-duplex at the voice message layerthat results in a kind of statistical
multiplexing of voice messages overthe WAN.
Nortel Communication Server 1000
IP TrunkFundamentals
NN43001-563 02.01 Standard
Release 5.5 21 December 2007
Copyright© 2007, Nor tel Networks
.