- This topic has 1 reply, 1 voice, and was last updated 25 years, 1 month ago by
.
-
Posts
-
Has anybody done any work on determining the IP bandwidth used when doing an VoIP call?
Even when the RTP, RDP, IP and Layer 2 overheads can be easily determined, depending on the type of voice codec (G.711, G.729A, G.723.1, etc) and the tools provided by the particular gateway you use (e.g. RTP header compression), the BW needed for one single call can vary considerably.
For example, using G.729A (8Kbps)in a Cisco 26XX can consume up to 26Kbps but if you can use RTP header compression (cisco proprietary) this can be reduced to aprox. 14-15Kbps, with a fixed number of voice frames per packet.
If you can adjust this quantity (cisco, lucent, etc), packing more frames per packet will reduce the impact of the OH and consequently improve the BW usage but you must look at the network latency for this bigger packets (this effects is strongly dependent of the network topology).
Other gateways (motorola, clarent)perform a sort of multiplexing at the IP level, packing voice frames from several incoming ports that go to the same remote gateway. This again reduce the impact of the OH and improve BW usage but take latency into account!
Best regards,
Guillermo Marmora.
- The forum ‘Telecom Design’ is closed to new topics and replies.
