Dustin.. WOW... This is what a good list serve is MADE OF... Love the detail.. Yes... Lemme know when you want to do the bake-off and lets try it out... Voip over Wireless Internet EVDO Bakeoff has the potential to get global sindication n'est pas??? :o)
X---------------- Robert Kim, Wireless Internet Wifi Hotspot Advisor http://wireless-internet-broadband-service.com https://evdo.sslpowered.com/wifi-hotspot-router.htm 2611 S Pacific Coast Highway 101 Cardiff by the Sea CA 92007 : 206 984 0880 >>> "Wireless Internet Service Is ONLY Broadband with Broadband Customer Service"(tm) >>> OUR QUEST: To Kill the Cubicle! (SM) ---Shalommmmmmmm-------------------- ---------------------------------;-)---- -----Original Message----- From: Dustin Goodwin [mailto:[EMAIL PROTECTED] Sent: Saturday, February 19, 2005 4:52 AM To: Robert Kim, EVDO-Coverage, Verizon Agent Cc: 'List Nyc Wireless' Subject: Re: [nycwireless] wireless internet modem bonding Great question... many things to consider. (Warning I don't have a simple answer) 1. Packet loss concealment. Several codecs on the market specialize in packet loss concealment. Global IP Sound has focused in on this space with their codecs. Skype I believe uses a GIS codec. http://www.skype.com/partners/current/. Check out slide 2 for comparison of different codecs with respect to packet loss levels. http://www.voipexperts.com/alcatel-fcc-tutorial-codec-slides.ppt 2. Latency. Total latency is something like: Sample size delay + encoding delay + network delay + jitter buffer + decoding delay. If we assume one way network delay of 120ms and our budget is 150ms (as most design guides suggest) we do not have much wiggle room. I just fired up my EVDO card and did some ping testing. Results: Ping statistics for 168.100.176.141: Packets: Sent = 28, Received = 28, Lost = 0 (0% loss), Approximate round trip times in milli-seconds: Minimum = 128ms, Maximum = 888ms, Average = 219ms My average round trip is 219ms with some awful swings up to almost 900ms and some excellent moments like128ms. Oh hell let's just call it 1 second. I noticed if I move my laptop at all I get packet loss which results in retransmission which results in extremely variable jitter :-(. So my little coffee table experiment yielded me a average one way delay of 110ms. Sounds like you have tested this stuff a lot more then me so I will stick with you 120ms number. As total latency rises above 150ms your calls start to get a ship to shore quality. The human brain has got telephone timing worked out pretty well. When you mess with the timing your brain throws a fit. In my own testing with a WAN simulator I have found 200ms is usable so long as neither party has had too much caffeine. Some things to consider in tuning latency: Codec processing delay Everything is a trade off with codec choice. Check out slide 1 in http://www.voipexperts.com/alcatel-fcc-tutorial-codec-slides.ppt Notice the higher bit rates codecs introduce the least latency. Of course this chart only covers ITU standard codecs. GIS and other specialist software companies can probably add other interesting codecs to the chart. Once again Skype with GIS codec probably excels in this space. Jitter buffer delay We don't have much choice in this matter. ITSPs usually configure their PSTN gateways with nominal jitters buffers then dynamically adjust according to actual under and over run conditions. Once again different gateway manufactures results in different dynamic jitter buffer performance. I hate to go back to GIS again but they seem to make the most noise in this space. Read http://www.globalipsound.com/datasheets/NetEQ.pdf (For the record I have no affiliation with GIS). 3. Tandem encoding. (Sounds like this might be your problem with calling cells phones) Check out slide 3 http://www.voipexperts.com/alcatel-fcc-tutorial-codec-slides.ppt And you will start to understand why calls to pstn might be fine but calls to cell phones suck. Some compression schemes just do not respond well to be placed in a back to back configuration. You may have to play with codec combinations to get optimal performance. We should do a bake off at one of the NYCwireless monthly meetings. Hookup a laptop with a few different clients/ITSP loaded and get a few testers with good ears in to judge audio quality to landlines and cell phones. You could bring your EVDO super blaster thingy to see if it helps. - Dustin - ps. The ppt slides I am using come from a tutorial given by Alcatel executive to the FCC's Office of Engineering and Technology. To get the entire preso go to http://www.fcc.gov/Bureaus/Engineering_Technology/Public_Notices/2001/d0 11369a.ppt . The real audio of the session is no longer online but if you send me an email I can point you to a mirror of the file. pss. This email sent over EVDO. Robert Kim, EVDO-Coverage, Verizon Agent wrote: >Dustin... Thanks for the positive word on EVDO... Noticed your VOIP >address... Do you know which codec is the best for EVDO's 120ms latency >issue...? > >I use EVDO for voip all the time.. But only to land lines... To cell >phones is just too difficult... Patchy... Ideas? > >Anyone? > >X---------------- >Robert Kim, >Wireless Internet Wifi Hotspot Advisor > >http://wireless-internet-broadband-service.com >https://evdo.sslpowered.com/wifi-hotspot-router.htm >2611 S Pacific Coast Highway 101 >Cardiff by the Sea CA 92007 : 206 984 0880 -- NYCwireless - http://www.nycwireless.net/ Un/Subscribe: http://lists.nycwireless.net/mailman/listinfo/nycwireless/ Archives: http://lists.nycwireless.net/pipermail/nycwireless/
