Silence suppression engineering considerations, Faxengineering considerations – Nortel Networks NN43001-563 User Manual

Page 114

Advertising
background image

114

ITG engineering guidelines

Codec

type

Codec
Multi -

frame

duration

(ms)

See Not

e

8.

Voice

/fax

paylo

ad

size

(byte

s)

IP

head

er

size

(byte

s)

Ether

net

head

er

size

(byte

s)

Full-d

uplex

Ether

net

Band

width

(bps)

PPP

WAN

Band

width

(bps)

See

Note

9.

Frame

Relay

WAN

bandw

idth

(bps)

ATM

WAN

bandw

idth

(bps)

10

10

40

26

60,800

45,600

46,400

84,800

20

20

40

26

34,400

26,800

27,200

42,400

DSP pro

fileAB/

G.729A

(8kbit/s)

voice

30

30

40

26

25,600

20,533

20,800

28,267

G.723.1

(5.3 kbit/s)

voice

30

20

40

26

22,933

17,867

18,133

26,571

G723.1

(6.3 kbit/s)

30

24

40

26

24,000

18,933

19,200

28,267

16.6

30

40

26

46,265

37,108

37,590

50,600

T.30/T.38

G3 Fax

25

30

40

26

30,720

24,960

24,960

33,900

Based on voice multiframe encapsulation for Realtime Transport Protocol per H.323 V2.

The bolded rows contain the default payload/packet size for each codec in TM 3.1.

TLAN subnet data rate is the effective Ethernet bandwidth consumption.

TLAN subnet kbit/s for voice traffic = 2*Ethernet frame bits*8/frame duration in ms

WAN kbit/s for voice traffic = IP packet bytes*8/frame duration in ms

Overhead (RTP/UDP header + IP header) of packets over the voice payload multiframe is 40
bytes; overhead of Ethernet frame over IP packet is 26 bytes.

An Interframe gap is not included in the above bandwidth calculation, because of the low probability
of occurring in this type of application.

Length of speech captured at each end. By definition, payload is one way.

These values do not include overhead from the network header (IEEE 802.3) that is automatically
added at the TLAN subnet link. To determine the approximate bandwidth used on the TLAN
subnet when including the network header, divide the values in the column "Bandwidth use on
TLAN subnet in kbit/s (two way)" by 2.

Silence Suppression engineering considerations

Silence Suppression/Voice Activity Detection (VAD) results in average
bandwidth savings over time, not in instantaneous bandwidth savings.
For normal conversations, Silence Suppression creates a 40% savings in
average bandwidth used. For example, a single G.729AB voice packet will
still consume 30 Kbps of bandwidth but the average bandwidth used for the
entire call would be approximately 23 Kbps.

Nortel Communication Server 1000

IP Trunk Fundamentals

NN43001-563

01.01

Standard

Release 5.0

30 May 2007

Copyright © 2007, Nortel Networks

.

Advertising