Other measurement considerations, Estimate voice quality – Nortel Networks NN43001-563 User Manual

Page 148

148

ITG engineering guidelines

In contrast, the site pair Santa Clara/Ottawa has both QoS levels of mean
and mean+standard deviation falling in the excellent region. The customer
has more confidence that during peak traffic period, the "excellent" service
level is likely to be upheld (better than 84% chance under the assumption of
Normal distribution).

Other measurement considerations

The PING statistics described above measure the intranet prior to IP Trunk
3.01 (and later) installation, which means that the measurement does not
take into consideration the expected load created by the IP Trunk 3.01 (and
later) users.

If the intranet capacity is tight and the IP Trunk 3.01 (and later) traffic
significant, consider making intranet measurements under load. Load can
be applied using traffic generator tools. The amount of load should match
the IP Trunk 3.01 (and later)-offered traffic estimated in

"IP Trunk 3.01 (and

later) traffic engineering" (page 95)

Estimate voice quality

The perceived quality of a telephone call is dependent on many factors,
such as codec characteristics, end-to-end delay, packet loss, and the
perception of the individual listener.

The E-Model Transmission Planning Tool is a model used to produce a
quantifiable measure of voice quality based on relevant factors. Refer to two
ITU-T recommendations (ITU-T E.107 and E.108) for more information
on the E-Model and its application.

A simplified version of the E-Model is applied to IP Trunk 3.01 (and later) to
provide an estimate of the voice quality that the user can expect, based on
various configuration choices and network performance metrics.

The simplified E-Model is as follows:

R = 94 – lc – ld – lp

where

= codec impairment (see

Table 29 "Impairment factors of codecs"

(page 149)

)

= delay impairment (see

Table 30 "Impairment factors due to network

delay" (page 149)

)

= packet loss impairment (see