Understanding the Requirements for Toll-Quality Voice Chapter 9: Network Requirements and Preparation
9.4.4 Jitter for Voice Switches
Jitter is the variation of latency across the network and the variation in packet processing
inside the switches. To compensate for jitter, the ShoreGear voice switches continuously
measure the jitter in the system and dynamically change the size of the receive jitter buffers
to optimize voice quality.
If the jitter buffer is too small, there can be packet loss from buffer underflows. This occurs
when the jitter buffer runs out of valid voice samples. If the jitter buffer is too large, there
will be unnecessary latency. Both conditions have a negative impact on voice quality.
The jitter buffer starts at the minimum size of 0 msecs as packets from the network are
placed into the switchboard queue for immediate processing. When jitter is detected on the
network, the jitter buffer dynamically increases in increments of 5 msecs to compensate for
increased jitter and decreases in size in reaction to less jitter. The maximum value of the
jitter buffer is set by ShoreWare Director and ranges from 20 to 300 msecs, with a default of
As the jitter increases on the network and the jitter buffer needs to be increased to
guarantee timely audio play, the latency of the audio also increases. The system attempts
both to maintain a minimum jitter buffer size that provides good-quality voice without
dropping packets and to provide minimum latency.
For IP phones that are configured into the ShoreTel system, the jitter buffer is not
configurable. The minimum jitter buffer is 10 msecs, and the maximum is 80 msecs.
Maximum values greater than 100 should rarely be necessary. If needed, this could indicate
a problem in your network that should be addressed in another way.
9.4.5 Packet Loss
Lost packets can occur on the IP network for any number of reasons. Packet loss above 1%
begins to adversely affect voice quality. To help reduce this problem, the ShoreGear voice
switches have a feature called lost packet concealment. When there is no voice sample to be
played, the last sample available is replayed to the receiving party at a reduced level. This is
repeated until a nominal level is reached, effectively reducing the clicking and popping
associated with low levels of packet loss.
Fax and modem calls demand essentially zero packet loss to avoid missing lines on fax calls
and to avoid dropped modem calls. In addition, fax and modem calls, when detected, may
change to a higher-rate codec.
LAN 17 5 5 Varies 5 32 + Jitter Buffer
WAN 17 5 15 Varies 5 42 + Jitter Buffer
(LAN and WAN)
17 15 15 Varies 15 62 + Jitter Buffer
a. The jitter buffer varies, depending on network conditions. See below for more informa-
b. If a call comes in on a trunk through either T1/E1 or analog loop-start, the total latency
is increased by the delay in the PSTN. You must add this latency to the total latency.
Latency for the PSTN varies; however, it is probably a minimum of 10 msecs (for local),
and it could be as high as hundreds of msecs (for long international calls).
Configuration Overhead Encoding Frame Size -5 Jitter BufferaDecoding Total (+/– 5 msec)b
Table 9-6 Latency