I am trying to make sure I can calculate bandwidth requirements for different vocoders correctly. Looking at G.711 with a sample size of 240, I get the right bandwidth for Frame Realy with and without cRTP. But ethernet gives me trouble. I am using the following method:
Total packet size (bytes) = (FrameRelay header of 6 bytes) + (IP/UDP/RTP header of 40 bytes) + (voice payload of 240 bytes) = 286 bytes
Total packet size (bits) = (240 bytes) * 8 bits per byte = 1920 bits
PPS = (64000 bps codec bit rate) / (1920 bits) = 33.3333 pps
Bandwidth per call = voice packet size (286 bits) * 33.33333 pps = 76267 Kbps
I get the right number of compressed traffic using cRTP. But when I try to calculate this for ethernet with a 18 byte header I get the wrong number. Would someone validate this? What do you get? Is my source incorrect? There is no errata for the text I am looking at. :-(
Message was edited by: sansari
Is the codec sample interval value ever going to change the result of this calculation? I am looking at the url: http://www.cisco.com/en/US/tech/tk652/tk698/technologies_tech_note09186a0080094ae2.shtml , and I am geting the same values. Seems Voice Payload Size is important; not the codec sample size in ms. I am just wondering if I have not run into an example where the sample size would change the packets per second. Seems sample size is a result of the speed of the coder.
