UMA over High-Latency Link
April 23, 2008 6:06 AM Subscribe
I've bought a new cell phone with UMA. Please tell me how it will behave over a satellite internet connection.
I have good WiFi reception in my apartment, but my cell service can be a little weak, so I've decided to get a phone with UMA. Most of the time I'll be using the WiFi over a cable modem connection, so no problems there. However, there will be times (2-3x/year for up to a week each) where I'll want to use the phone in the boonies. At this location we have WiFi through a satellite ISP. This means ping times in the neighborhood of one second (1000ms), but the throughput is good (1Mbps/256kbps down/up). The sales rep was unsure, and this is not a necessary feature, so I didn't push, but it would be nice if others in the hive mind have had experience.
I have good WiFi reception in my apartment, but my cell service can be a little weak, so I've decided to get a phone with UMA. Most of the time I'll be using the WiFi over a cable modem connection, so no problems there. However, there will be times (2-3x/year for up to a week each) where I'll want to use the phone in the boonies. At this location we have WiFi through a satellite ISP. This means ping times in the neighborhood of one second (1000ms), but the throughput is good (1Mbps/256kbps down/up). The sales rep was unsure, and this is not a necessary feature, so I didn't push, but it would be nice if others in the hive mind have had experience.
Response by poster: b1tr0t, I realize that the delay will be wicked annoying for voice calls, but it shouldn't matter for things like MMS and SMS, which is what I'd primarily be using way out there.
cgomez, thanks for the warning about even the latencies found on cable connections. I will be in contact with my cable provider about caps and about improving our currently iffy latency performance.
posted by Xoder at 6:25 AM on April 23, 2008
cgomez, thanks for the warning about even the latencies found on cable connections. I will be in contact with my cable provider about caps and about improving our currently iffy latency performance.
posted by Xoder at 6:25 AM on April 23, 2008
I think that UMA phones take round-trip latency into account in determining the relative channel quality of the UMA/Wifi connection. So, if there is even the faintest shred of GSM signal available in your boondocks location, you may need to set the phone to "UMA-only" or "WiFi-only" mode in order to get it to stay on your satellite connection.
posted by Juffo-Wup at 6:31 AM on April 23, 2008
posted by Juffo-Wup at 6:31 AM on April 23, 2008
I've used VOIP over my HughesNet connection. Apart from having to pretend you're on the moon (including saying "over" when you're done talking), it works pretty good.
I've done a few recent experiments with video conferencing, too. Some programs with high quality requirements, such as iChat, don't work at all, but I found Skype to be surprisingly useful.
And, as mentioned above, HughesNet's bandwidth caps are brutal. It sounds like you may have the ProPlus plan, so you should be generally ok with voice traffic.
posted by ewagoner at 6:32 AM on April 23, 2008
I've done a few recent experiments with video conferencing, too. Some programs with high quality requirements, such as iChat, don't work at all, but I found Skype to be surprisingly useful.
And, as mentioned above, HughesNet's bandwidth caps are brutal. It sounds like you may have the ProPlus plan, so you should be generally ok with voice traffic.
posted by ewagoner at 6:32 AM on April 23, 2008
Response by poster: I do not know which satellite ISP the place I go to uses, but they probably have an unlimited version, as they also run a summer camp there, and caps would be bad for the campers...
posted by Xoder at 8:41 AM on April 23, 2008
posted by Xoder at 8:41 AM on April 23, 2008
I realize that the delay will be wicked annoying for voice calls, but it shouldn't matter for things like MMS and SMS, which is what I'd primarily be using way out there.
I wouldn't expect those to work over WiFi at all, but it looks like I'm wrong and they should.
posted by kindall at 10:23 AM on April 23, 2008
I wouldn't expect those to work over WiFi at all, but it looks like I'm wrong and they should.
posted by kindall at 10:23 AM on April 23, 2008
[i]I wouldn't expect those to work over WiFi at all, but it looks like I'm wrong and they should.[/i]
Those things don't work on all WiFi-enabled cell phones, only ones with UMA (aka GAN). UMA/GAN works by building an encrypted tunnel over the Internet back to one of the switching centers operated by the subscriber's cell phone carrier. The phone then treats this tunnel as another candidate for network access, just like any nearby cell towers. So, anything that works over a normal cell connection works over UMA.
It's sort of a niche technology but it's great in situations like Xoder's and for people who have poor cellular coverage at their homes/offices/etc.
posted by Juffo-Wup at 6:51 PM on April 23, 2008
Those things don't work on all WiFi-enabled cell phones, only ones with UMA (aka GAN). UMA/GAN works by building an encrypted tunnel over the Internet back to one of the switching centers operated by the subscriber's cell phone carrier. The phone then treats this tunnel as another candidate for network access, just like any nearby cell towers. So, anything that works over a normal cell connection works over UMA.
It's sort of a niche technology but it's great in situations like Xoder's and for people who have poor cellular coverage at their homes/offices/etc.
posted by Juffo-Wup at 6:51 PM on April 23, 2008
Best answer: With the help of some fine folks in #mefi, I was able to set up my router to add a specified amount of delay to packets coming in and out of my network.
With this in place, I did some testing of my T-Mobile UMA phone and found that it was able to maintain a UMA connection and use it for voice and data with round trip delay times of up to approximately 2100ms. Anything higher than that and the phone was unable to maintain a connection to the network via UMA.
I found that I had to set the phone to WiFi-only mode to get it to stay on the WiFi with round trip times of over 250ms or thereabouts. But, I was testing this in an area with a relatively strong (-70 dBm) cellular signal, so you might not have to bother if you are someplace with weak or no cell signal.
Other than the annoyance of the actual mouth-to-ear delay, the increased latency did not seem to cause problems with the audio quality during calls. Some of the calls I made were to automated menu systems, and some of the DTMF digits I dialed were not recognized by the menu system on the longer delay calls. I don't know exactly why this is, but I wouldn't plan on being able to enter your 16 digit credit card number (or what have you) while using the satellite connection.
Finally, although I induced substantial delays into my connection, I did not induce any packet loss or jitter. Your satellite connection is likely to have a substantial amount of jitter (I don't know about packet loss), and I'm not sure what type of jitter buffering algorithms are used in UMA. So it's possible that excess jitter or packet loss could cause call problems for you, but it appears that the latency itself shouldn't prevent you from connecting or placing calls.
posted by Juffo-Wup at 7:58 PM on April 23, 2008
With this in place, I did some testing of my T-Mobile UMA phone and found that it was able to maintain a UMA connection and use it for voice and data with round trip delay times of up to approximately 2100ms. Anything higher than that and the phone was unable to maintain a connection to the network via UMA.
I found that I had to set the phone to WiFi-only mode to get it to stay on the WiFi with round trip times of over 250ms or thereabouts. But, I was testing this in an area with a relatively strong (-70 dBm) cellular signal, so you might not have to bother if you are someplace with weak or no cell signal.
Other than the annoyance of the actual mouth-to-ear delay, the increased latency did not seem to cause problems with the audio quality during calls. Some of the calls I made were to automated menu systems, and some of the DTMF digits I dialed were not recognized by the menu system on the longer delay calls. I don't know exactly why this is, but I wouldn't plan on being able to enter your 16 digit credit card number (or what have you) while using the satellite connection.
Finally, although I induced substantial delays into my connection, I did not induce any packet loss or jitter. Your satellite connection is likely to have a substantial amount of jitter (I don't know about packet loss), and I'm not sure what type of jitter buffering algorithms are used in UMA. So it's possible that excess jitter or packet loss could cause call problems for you, but it appears that the latency itself shouldn't prevent you from connecting or placing calls.
posted by Juffo-Wup at 7:58 PM on April 23, 2008
Response by poster: Wow, Juffo-Wup, you have been most excellent! That answers the last of my concerns about this new phone, and I'm sure I'll love it whenever someone's at my place in the afternoon when UPS comes...
In return, I shall give you "Best Answer" and explain to you why DTMF is handled differently than regular voice traffic!
See, the audio compression algorithms that GSM uses are tuned to voice very well (i.e. low-latency, extremely high compression), not tones. So the GSM protocol has a trick: There are some overhead time-slots left over every super-epoch where it puts control data (things like handover notices). If you hit the buttons, it will send a code during these time-slots instead of trying to force it over the audio channel. The tone is generated before entering the public telephone network.*
Thanks again for the info, and for the heroic router modifications!
* However, I do not know where the tone is generated if the call never leaves the cell provider's internal switched network. It may even be generated at the remote phone...
posted by Xoder at 8:21 PM on April 23, 2008
In return, I shall give you "Best Answer" and explain to you why DTMF is handled differently than regular voice traffic!
See, the audio compression algorithms that GSM uses are tuned to voice very well (i.e. low-latency, extremely high compression), not tones. So the GSM protocol has a trick: There are some overhead time-slots left over every super-epoch where it puts control data (things like handover notices). If you hit the buttons, it will send a code during these time-slots instead of trying to force it over the audio channel. The tone is generated before entering the public telephone network.*
Thanks again for the info, and for the heroic router modifications!
* However, I do not know where the tone is generated if the call never leaves the cell provider's internal switched network. It may even be generated at the remote phone...
posted by Xoder at 8:21 PM on April 23, 2008
Response by poster: Just wanted to leave the following update: My phone worked great over Satellite Internet for both MMS and SMS! The connection had too much jitter for voice to work, but I wasn't planning to use it for that, so it was OK.
Side Note: Thanks to the iffy latency and throughput from my cable provider, I also stopped using WiFi from my apartment, as calling people and sounding like, "Burwhap fjanns, oowhe?" was not acceptable.
posted by Xoder at 3:52 PM on October 13, 2008
Side Note: Thanks to the iffy latency and throughput from my cable provider, I also stopped using WiFi from my apartment, as calling people and sounding like, "Burwhap fjanns, oowhe?" was not acceptable.
posted by Xoder at 3:52 PM on October 13, 2008
This thread is closed to new comments.
posted by cgomez at 6:09 AM on April 23, 2008