Following up on why EDGE versus 3G matters less than you think
After being Slashdotted, yesterday's article regarding EDGE and 3G certainly stirred up a lot of commenting and controversy. I thought I would try to clarify some points today that came up in comments on the article.
First and foremost, I want to point everyone to the title of the article: Why EDGE versus 3G may not matter as much as you think. Despite comments to the contrary, I didn't claim that EDGE was better than 3G; I simply said it didn't matter as much as most people thought. Unless we agree on that initial point, further argument probably won't convince anyone of anything. And I completely agree that EDGE is quite slow from a bandwidth perspective. It is. Really.
What I did say, though, is that bandwidth doesn't matter as much as most people think. Specifically, I made the point that user experience is going to depend just as much on 1) latency, 2) error rates, 3) phone processor speed, and 4) power consumption. And not all of these characteristics improve with higher speed networks. In fact, at least one of them -- power consumption -- gets worse. And to the commenters who questioned whether error rates do in fact go up with higher bandwidth, that question a good one, but too complex to discuss here; a good starting point for discussion is Claude Shannon's Mathematical Theory of Communication, but that's just a beginning.
One of the issues that makes this a challenging topic to write about is that while there is only one EDGE, there are many 3G networks and layers, including UMTS, W-CDMA, CDMA2000, EVDO and others. Further, all those alphabet soup standards have real world implementation warts when they get actually built. Some networks have great latency characteristics, and others have not so great ones ones. So it's necessary to be specific about the comparisons.
My understanding is, in fact, that UMTS networks specifically have much better latency than EDGE (I don't have a UMTS network here to play with on a daily basis, but that's what I've read). However, they are also quite wideband technologies, using 5 MHz channels, which has meant that US carriers have had trouble getting spectrum for them, and hence, we don't see a ton of UMTS here, except in major urban areas where carriers have been able to buy additional spectrum. We do, however, have some HSPDA (which, to be clear, is a type of 3.5G version of UMTS) in major cities like New York, and on those networks, latency runs anywhere from 150 to 250 milliseconds. To put that in perspective, the average WiFi network provides latencies on the order of 1 to 10 milliseconds. One commenter did point out, though, that I haven't measured the latency of the EDGE network. Your point is well-taken, and I intend to do so soon.
I also have no direct experience with the Verizon EVDO network since I'm not a subscriber, so I can't compare its latency or error characteristics. Overall, I"m a big fan of spread spectrum systems because they manage to multiplex data streams in much better ways. That said, I'm a GSM user because that's the dominant standard and it's not going to change any time soon.
However, the most important point I want to make is the one I made in yesterday's article and has nothing to do with technology: obsessing about 3G bandwidth in the absence of consumer use and applications makes no sense. The reality is that ordinary consumers use very few high-bandwidth applications of any kind for regular communication. Even the high speed video streaming pushed by the likes of Verizon and ATT is of no interest to 75% of consumers. Instead, what consumers do use is a lot of applications that demand an always-on, fairly low-latency network. Examples of that include phone calls, email, instant messaging, Web browsing, VOIP, Google maps, you name it. Any company, not just Apple, who can figure out how to deliver great experiences on low-bandwidth networks, has a huge advantage over those that require 3G because 1) they have a much larger addressable market, and 2) their service can get used in more places for years to come. And no matter how much we may argue about the various technical merits, that market reality isn't likely to change within this decade.
First and foremost, I want to point everyone to the title of the article: Why EDGE versus 3G may not matter as much as you think. Despite comments to the contrary, I didn't claim that EDGE was better than 3G; I simply said it didn't matter as much as most people thought. Unless we agree on that initial point, further argument probably won't convince anyone of anything. And I completely agree that EDGE is quite slow from a bandwidth perspective. It is. Really.
What I did say, though, is that bandwidth doesn't matter as much as most people think. Specifically, I made the point that user experience is going to depend just as much on 1) latency, 2) error rates, 3) phone processor speed, and 4) power consumption. And not all of these characteristics improve with higher speed networks. In fact, at least one of them -- power consumption -- gets worse. And to the commenters who questioned whether error rates do in fact go up with higher bandwidth, that question a good one, but too complex to discuss here; a good starting point for discussion is Claude Shannon's Mathematical Theory of Communication, but that's just a beginning.
One of the issues that makes this a challenging topic to write about is that while there is only one EDGE, there are many 3G networks and layers, including UMTS, W-CDMA, CDMA2000, EVDO and others. Further, all those alphabet soup standards have real world implementation warts when they get actually built. Some networks have great latency characteristics, and others have not so great ones ones. So it's necessary to be specific about the comparisons.
My understanding is, in fact, that UMTS networks specifically have much better latency than EDGE (I don't have a UMTS network here to play with on a daily basis, but that's what I've read). However, they are also quite wideband technologies, using 5 MHz channels, which has meant that US carriers have had trouble getting spectrum for them, and hence, we don't see a ton of UMTS here, except in major urban areas where carriers have been able to buy additional spectrum. We do, however, have some HSPDA (which, to be clear, is a type of 3.5G version of UMTS) in major cities like New York, and on those networks, latency runs anywhere from 150 to 250 milliseconds. To put that in perspective, the average WiFi network provides latencies on the order of 1 to 10 milliseconds. One commenter did point out, though, that I haven't measured the latency of the EDGE network. Your point is well-taken, and I intend to do so soon.
I also have no direct experience with the Verizon EVDO network since I'm not a subscriber, so I can't compare its latency or error characteristics. Overall, I"m a big fan of spread spectrum systems because they manage to multiplex data streams in much better ways. That said, I'm a GSM user because that's the dominant standard and it's not going to change any time soon.
However, the most important point I want to make is the one I made in yesterday's article and has nothing to do with technology: obsessing about 3G bandwidth in the absence of consumer use and applications makes no sense. The reality is that ordinary consumers use very few high-bandwidth applications of any kind for regular communication. Even the high speed video streaming pushed by the likes of Verizon and ATT is of no interest to 75% of consumers. Instead, what consumers do use is a lot of applications that demand an always-on, fairly low-latency network. Examples of that include phone calls, email, instant messaging, Web browsing, VOIP, Google maps, you name it. Any company, not just Apple, who can figure out how to deliver great experiences on low-bandwidth networks, has a huge advantage over those that require 3G because 1) they have a much larger addressable market, and 2) their service can get used in more places for years to come. And no matter how much we may argue about the various technical merits, that market reality isn't likely to change within this decade.
Technorati Tags: Bandwidth, EDGE, Latency, Mobile phone, Mobile phones, UMTS, W-CDMA
posted by Carl Howe at
8:59 AM
| View blog reactions
| 
