NEC SL1100, Сетевое руководство

Страница: 75 / 292

NEC SL1100 Issue 6.0
Networking Manual 6 - 13
5.2 Bandwidth
The bandwidth required for VoIP calls depends on several factors, including:

Number of simultaneous calls

CODEC used

Frame Size

Data Networking Protocol used
The more frames encapsulated into each packet, the less bandwidth is
required. This is because each packet transmitted has the same header size.
Therefore, if numerous very small packets are sent then bandwidth is also
being used for a large amount of header information. If we add several frames
to the packet, less packets are transmitted and therefore have less header
information sent.
If we add many voice frames to each packet, less bandwidth is being used.
However, this does have disadvantages. If there is a large packet size, and a
particular voice packet is lost, this has a greater impact on the speech quality.
If a small quantity of voice frames per packet is being used, the effect of losing
a packet is reduced.
As a general rule: The more frames per packet, the less bandwidth is used,
but the quality is also lower.
Examples:
Example 1: CODEC: G.729 Frame Size: 10ms Voice Frames per Packet: 2
Voice Sample Size: 20ms (frame size x Voice Frames) Bandwidth Required:
24Kbps
Example 2: CODEC: G.729 Frame Size: 80ms Voice Frames per Packet: 8
Voice Sample Size: 80ms (frame size x Voice Frames) Bandwidth Required:
12Kbps
S
ECTION
6
Q
UALITY OF
S
ERVICE
(Q
O
S) I
MPLEMENTATION
Section 2.2 Voice Quality Improvements on page 6-3 discusses some of the problems
associated with voice quality. This section describes how QoS can be implemented
on data networks to provide the “best case” for VoIP traffic.