A couple of quick comments, based on recent discussions:
1) Some people in the US are using the term "aircard" (Sierra Wireless' brand) as a generic term for 3G/WiMAX modem, rather than "dongle"
2) Some people (also mostly from the US or Canada) are still under the impression that most mobile broadband users are corporate "road warriors", and have very limited awareness of the huge massmarket of consumer dongle buyers around the world.
3) The phenomenon of prepaid mobile broadband remains quite low in awareness among people "in the industry"
4) I heard of a European operator subsidising *unlocked* embedded-3G modules in notebooks sold through *retail* channels. In other words, you open the box, get a note saying something like "Congratulations! Your laptop has 3G! We're giving you a SIM and a month's access for free...." but you can still subsequently swap it and use a competitor's SIM if you decide to, after the trial. (This is unconfirmed - but it sounds to me like an extremely expensive version of AOL's old CD-in-the-PC-box promotion technique from the 1990s)
Report: Mobile Broadband Computing
Thursday, June 04, 2009
Dissecting mobile broadband stats
I noticed that the other day that the GSMA put out a press release stating
"Today, there are 245 commercially available HSPA networks supporting more than 125 million live connections in 107 countries. A further 65 HSPA networks are either being deployed, in trial or planned with an average of four million Mobile Broadband connections being added globally on a monthly basis. "
I followed up to get a better definition of what's in that 125m - it turns out that it's a combination of 3G SIM + HSPA device, but does not reflect actual usage. So for example, today I'm accounting for 3 of those connections, with my ZTE HSPA dongle, my Nokia E71 with data SIM, and my SonyEricsson C902 (which is permanently switched to 2G-only for extended battery-life).
I'm wondering what the make-up of that 125m is, in more detail, in a way which suggests actual usage.
My quick and very rough estimates are:
Laptop dongles & embedded modems (actively used) - 35m
Laptop dongles & embedded modems (dormant or inactive) - 5m
3G iPhones (with active data usage, NB most have flatrate data) - 20m
Other HSPA smartphones (active HSPA data usage with flatrate or "decent" data plans) - 20m
Other HSPA smartphones (dormant or without decent data plans / 3G SIM) - 35m
HSPA featurephones (mostly minimal data usage) - 10m
I'm fairly confident about the active laptop use numbers as they originate in my Mobile Broadband Computing research report from a few months ago. I reckon that virtually all 3G iPhones are used "in anger" with HSPA.
But the bulk of other HSPA-enabled smartphones are not used "aggressively", as unlike the Apple they are not always sold with the flatrate data plan as a default option. There's an awful lot of older Nokia N95's, HTC WinMob handsets and similar devices around, used without heavy (or any) data consumption. Newer models like E71's, or Android G1's or BlackBerry Bolds are more-used for data, albeit with lower consumption than most iPhones.
(One variable I'm not too sure about is exactly how to categorise the DoCoMo and Softbank HSPA phones).
Thoughts and comments welcome - this is really just a first pass.
"Today, there are 245 commercially available HSPA networks supporting more than 125 million live connections in 107 countries. A further 65 HSPA networks are either being deployed, in trial or planned with an average of four million Mobile Broadband connections being added globally on a monthly basis. "
I followed up to get a better definition of what's in that 125m - it turns out that it's a combination of 3G SIM + HSPA device, but does not reflect actual usage. So for example, today I'm accounting for 3 of those connections, with my ZTE HSPA dongle, my Nokia E71 with data SIM, and my SonyEricsson C902 (which is permanently switched to 2G-only for extended battery-life).
I'm wondering what the make-up of that 125m is, in more detail, in a way which suggests actual usage.
My quick and very rough estimates are:
Laptop dongles & embedded modems (actively used) - 35m
Laptop dongles & embedded modems (dormant or inactive) - 5m
3G iPhones (with active data usage, NB most have flatrate data) - 20m
Other HSPA smartphones (active HSPA data usage with flatrate or "decent" data plans) - 20m
Other HSPA smartphones (dormant or without decent data plans / 3G SIM) - 35m
HSPA featurephones (mostly minimal data usage) - 10m
I'm fairly confident about the active laptop use numbers as they originate in my Mobile Broadband Computing research report from a few months ago. I reckon that virtually all 3G iPhones are used "in anger" with HSPA.
But the bulk of other HSPA-enabled smartphones are not used "aggressively", as unlike the Apple they are not always sold with the flatrate data plan as a default option. There's an awful lot of older Nokia N95's, HTC WinMob handsets and similar devices around, used without heavy (or any) data consumption. Newer models like E71's, or Android G1's or BlackBerry Bolds are more-used for data, albeit with lower consumption than most iPhones.
(One variable I'm not too sure about is exactly how to categorise the DoCoMo and Softbank HSPA phones).
Thoughts and comments welcome - this is really just a first pass.
Wednesday, June 03, 2009
WiMAX and LTE - a tale of two conferences
I've been in Amsterdam with the WiMAX Forum for the past couple of days, at their global summit event. It's been an interesting contrast with the LTE Summit in Berlin a fortnight ago.
In format terms, the LTE event was more of a traditional conference, with bigger plenary sessions. The WiMAX event was more of a trade show, with more exhibition visitors than delegates to the speaker sessions.
The LTE event felt as though the wheels of hype were starting to squeak. There was a palpable feeling that the timelines are overenthusiastic, the wrangling of voice dominated discussions, and there was quite a lot of dissent about business models and how LTE would coexist with/replace HSPA. LTE's definitely still "work in progress" and is unlikely to be deployed in anger outside the US and Japan for some considerable time. Many operators will go via HSPA+ in the short term instead.
Conversely, the WiMAX event showed more signs of positivity I expected. I've long joked that WiMAX goes through a roughly 4-month cycle of oscillating optimism and pessimism - but it seems to hold true.
Although there was still some rhetoric from the Forum, and Intel and cheerleaders like Clearwire presenting WiMAX as some radical high-performance alternative to HSPA, or "4G before LTE is nearly ready", the reality is pretty easy to discern.
Firstly, WiMAX opportunity is largely determined by spectrum availability and regulation. There's still no likelihood of commercial FDD versions of WiMAX in the near future, so the key story is that WiMAX is the technology of choice for any provider that has access to TDD spectrum.
The most interesting thing for me was the upswing of 2.3GHz, rather than just the normal 2.5GHz band that most expect to be the core domain for WiMAX. Coupled with dual- and tri-band equipment (and roaming starting to be offered), this changes the game somewhat. I'm not aware of any HSPA or LTE silicon provide seriously looking at 2.3GHz as an important band, so in that part of the spectrum, WiMAX is pretty much the only game in town. Given that LTE-TDD is even less developed than LTE-FDD, those countries that have licenced 2.5GHz bands are also likely to get at least one WiMAX provider.
There is still a lot of emphasis on using WiMAX as an alternative to DSL in markets that don't have much copper. Given the relatively cheap price of TDD spectrum, it seems that WiMAX is easier to justify than fixed-3G routers. Although numerous companies have HSPA routers, there doesn't appear to have been a huge amount of traction in the market - possibly because of the load that heavy users place on the cells.
My general belief is that mobility-optimised networks like HSPA are too complex and expensive to be wasted on non-mobile users. Obviously this will vary somewhat, but seems to suggest to me that fixed-HSPA and fixed-LTE deployments will struggle in many instances.
One interesting observation has been around backhaul and network dimensioning. Because WiMAX operators are starting from a base of fixed-CPE and high-end nomadic usage, they seem to be anticipating much higher loads. A central planning assumption seems to be that an "average" user could well be using 5-10GB per month, and this is being reflected in pricing plans and also use of high-speed backhaul.
Some other quick notes:
- most WiMAX operators are talking about access to "the real Internet", with no particular messing-about with DPI or application filtering
- there was a fair amount of talk about VoIP, but it's generally intended more for fixed/nomadic usage than true mobility. My assumption is that everyone with a WiMAX device for the next 5 years will also have a separate GSM phone - or else will have a dual-mode WiMAX/GSM handset anyway.
- Lots of interesting stuff happening in Russia, where 3G deployment has been slowed for various reasons (eg military use of spectrum). Comstar, Yota & Enforta had interesting pitches - with Yota being especially aggressive.
- Lots of talk about netbooks, and embedded-WiMAX PCs. It seems likely that embedding modules will be country-specific, perhaps with predominantly WiMAX modules in markets like Russia, but HSPA in markets like Scandinavia. Medium-term, I think a fairly high % of laptops will need to be dual-mode HSPA (or LTE) plus WiMAX, which should pose some interesting challenges for the connection management software.
- There should be some interesting business models emerging, with prepay from Day 1 in many markets. It's still early days for adhoc usage, though
- The UQ proposition in Japan sounds interesting, especially given its part-ownership by KDDI. It looks like it's aiming to be a true, low-cost "pipe" provider, which sounds like a serious differentiator in a market always dominated by operator-managed services for mobile devices.
- I was unconvinced that WiMAX has any chance to compete with cheap massmarket HSPA dongles in Europe - not because of any specific failings, but because bargain-basement HSPA pricing seems to be lossmaking for many operators at the moment, priced at less than the cost-of-production per GB.
Overall, I still don't think that WiMAX is a real "competitor" for LTE. Ultimately, it will probably only address 20-30% of the accessible mobile broadband spectrum in most countries. It still faces challenges getting non-PC/dongle devices to market in sufficient quantities - I'm skeptical we'll see a WiMAX iPhone any time soon. But it fulfills a couple of important roles in various markets, notably for "wireless DSL", and mobile broadband in markets for which HSPA dongles are either unavailable or very expensive (eg US).
In format terms, the LTE event was more of a traditional conference, with bigger plenary sessions. The WiMAX event was more of a trade show, with more exhibition visitors than delegates to the speaker sessions.
The LTE event felt as though the wheels of hype were starting to squeak. There was a palpable feeling that the timelines are overenthusiastic, the wrangling of voice dominated discussions, and there was quite a lot of dissent about business models and how LTE would coexist with/replace HSPA. LTE's definitely still "work in progress" and is unlikely to be deployed in anger outside the US and Japan for some considerable time. Many operators will go via HSPA+ in the short term instead.
Conversely, the WiMAX event showed more signs of positivity I expected. I've long joked that WiMAX goes through a roughly 4-month cycle of oscillating optimism and pessimism - but it seems to hold true.
Although there was still some rhetoric from the Forum, and Intel and cheerleaders like Clearwire presenting WiMAX as some radical high-performance alternative to HSPA, or "4G before LTE is nearly ready", the reality is pretty easy to discern.
Firstly, WiMAX opportunity is largely determined by spectrum availability and regulation. There's still no likelihood of commercial FDD versions of WiMAX in the near future, so the key story is that WiMAX is the technology of choice for any provider that has access to TDD spectrum.
The most interesting thing for me was the upswing of 2.3GHz, rather than just the normal 2.5GHz band that most expect to be the core domain for WiMAX. Coupled with dual- and tri-band equipment (and roaming starting to be offered), this changes the game somewhat. I'm not aware of any HSPA or LTE silicon provide seriously looking at 2.3GHz as an important band, so in that part of the spectrum, WiMAX is pretty much the only game in town. Given that LTE-TDD is even less developed than LTE-FDD, those countries that have licenced 2.5GHz bands are also likely to get at least one WiMAX provider.
There is still a lot of emphasis on using WiMAX as an alternative to DSL in markets that don't have much copper. Given the relatively cheap price of TDD spectrum, it seems that WiMAX is easier to justify than fixed-3G routers. Although numerous companies have HSPA routers, there doesn't appear to have been a huge amount of traction in the market - possibly because of the load that heavy users place on the cells.
My general belief is that mobility-optimised networks like HSPA are too complex and expensive to be wasted on non-mobile users. Obviously this will vary somewhat, but seems to suggest to me that fixed-HSPA and fixed-LTE deployments will struggle in many instances.
One interesting observation has been around backhaul and network dimensioning. Because WiMAX operators are starting from a base of fixed-CPE and high-end nomadic usage, they seem to be anticipating much higher loads. A central planning assumption seems to be that an "average" user could well be using 5-10GB per month, and this is being reflected in pricing plans and also use of high-speed backhaul.
Some other quick notes:
- most WiMAX operators are talking about access to "the real Internet", with no particular messing-about with DPI or application filtering
- there was a fair amount of talk about VoIP, but it's generally intended more for fixed/nomadic usage than true mobility. My assumption is that everyone with a WiMAX device for the next 5 years will also have a separate GSM phone - or else will have a dual-mode WiMAX/GSM handset anyway.
- Lots of interesting stuff happening in Russia, where 3G deployment has been slowed for various reasons (eg military use of spectrum). Comstar, Yota & Enforta had interesting pitches - with Yota being especially aggressive.
- Lots of talk about netbooks, and embedded-WiMAX PCs. It seems likely that embedding modules will be country-specific, perhaps with predominantly WiMAX modules in markets like Russia, but HSPA in markets like Scandinavia. Medium-term, I think a fairly high % of laptops will need to be dual-mode HSPA (or LTE) plus WiMAX, which should pose some interesting challenges for the connection management software.
- There should be some interesting business models emerging, with prepay from Day 1 in many markets. It's still early days for adhoc usage, though
- The UQ proposition in Japan sounds interesting, especially given its part-ownership by KDDI. It looks like it's aiming to be a true, low-cost "pipe" provider, which sounds like a serious differentiator in a market always dominated by operator-managed services for mobile devices.
- I was unconvinced that WiMAX has any chance to compete with cheap massmarket HSPA dongles in Europe - not because of any specific failings, but because bargain-basement HSPA pricing seems to be lossmaking for many operators at the moment, priced at less than the cost-of-production per GB.
Overall, I still don't think that WiMAX is a real "competitor" for LTE. Ultimately, it will probably only address 20-30% of the accessible mobile broadband spectrum in most countries. It still faces challenges getting non-PC/dongle devices to market in sufficient quantities - I'm skeptical we'll see a WiMAX iPhone any time soon. But it fulfills a couple of important roles in various markets, notably for "wireless DSL", and mobile broadband in markets for which HSPA dongles are either unavailable or very expensive (eg US).
Tuesday, June 02, 2009
Individual bandwidth management based on radio conditions
Interesting comment from a Danish WiMAX operator I heard at an event yesterday - they actually decide on application throttling / bandwidth management based on individual users' radio channel conditions.
The logic is that if you're in a strong signal area, the modulation is such that you need a smaller % of total base station capacity, compared to a user at the edge of the cell, trying to pull down the same amount of bandwidth.
In other words, they're basing policy decisions on actual capacity & resource usage, not using data transfer rates as a proxy.
Presumably, this needs some sort of realtime API into the air interface part of the network - exactly the sort of thing I had in mind a couple of weeks ago when I asked the LTE folk about programmable networks.
The logic is that if you're in a strong signal area, the modulation is such that you need a smaller % of total base station capacity, compared to a user at the edge of the cell, trying to pull down the same amount of bandwidth.
In other words, they're basing policy decisions on actual capacity & resource usage, not using data transfer rates as a proxy.
Presumably, this needs some sort of realtime API into the air interface part of the network - exactly the sort of thing I had in mind a couple of weeks ago when I asked the LTE folk about programmable networks.
Subscribe to:
Posts (Atom)
Posts
