Wireless telephony is undergoing a revolution, with technology and implementation philosophy each holding the other back in turn as the industry struggles towards wireless nirvana. Engineers have spent the last few decades squeezing more data into the same wireless bandwidth, a process that is deep into the realm of diminishing …
Let me be the first to say...
Fantastic article, Bill. Top notch.
Stuff like this is why I read El Reg. Well this and hoping one day to find a competition in which the prize is Ms. Bee banging on my door with Twister, baby oil and a cheeky grin...
A history lesson
"That changed with GPRS (General Packet Radio Service), which did at least use a digital connection..."
Errm - funnily enough, CSD over GSM also use[s/d]* a digital connection too. GSM doesn't do analogue - unlike its predecessors.
*delete as applicable.
Missing factor - idiotic data plans
I know this article was about the underlying technology but it seems that a little time could have been spent discussing the real reason that mobile data is so underused - idiotic data plans. Network operators continually fucked up the usability of mobile data by offering data plans that meant most people couldn't know what it would cost them to visit a website or download an app, but knew that it probably wasn't low enough to write it off. Too much worry about cost made it not worth doing for the majority.
Operators did virtually nothing to address that until very recently (the arrival of the iPhone with its all-you-can-eat data bundle is an example of how to make it easy to predict).
Good but missing Gotchas
an In depth comment
These speeds are TOTAL PEAK Capacity for sector and only achievable close to a mast. More aerials adds serious costs (as it needs a shelf of electronics) and only increases nearby capacity, not peak user speed. The 160Mbps or 300Mbps translates to a 4Mbps to 5Mbps total throughput a virtual sector edge. Or about 1Mbps per person. Only just entry level Broadband.
The future is LTE not WiMax.
... and for good voice coverage Operators may do better to keep GSM than switch to "breathing" W-CDMA based 3G with its unreliable coverage.
It will be 4 to 10 time better than 3G. But coverage and speed/capacity depends on Cell density. MIMO adds a lot of cost for extra capacity. More ROI by not deploying MIMO and simply having x2 to x4 slower speeds for same capacity. (peak speeds are the same, with or without MIMO).
Overall a fairly good article, though lacking the "gotchas".
The Mobile Industry still seems to suffer from delusions. They should make the top execs use HSPA and LTE modems exclusively on a economically loaded mast (for ROI) and then on a decent Fibre fed Cable and/or FTTC VDSL/ADSL-2 and then they might understand the difference between Mobile and Broadband.
The Mobile Operators should help companies put modems in Gadgets (money, subsidised operator sales etc). Play to their strength of Mobility, not compete with fixed Broadband.
Also we need a level palying field. LTE is NOT going to be cheap. But more expensive than GSM/3G as it the subsidy by voice can't continue. LTE is also VOIP only, it's not a telephone system with Data tacked on as GSM/3G is.
See me after class.
I can't really fault the general direction of this article, but if you are going to sprinkle it with technical info then you really should get it right.
CSD and GPRS are both digital. CSD does not convert data into audible tones. The difference between CSD and GPRS is that one is circuit switched and one is packet switched. The clue is in the name.
EDGE is categorically not a software upgrade, it uses an entirely different modulation scheme.
I put on my pedant's mortarboard at this point and clicked the comment button. I could probably be equally tedious about the rest of the article for the price of a pint.
So does that mean WiMax is a no-go? I'm presuming we'll all be using iPhone 4G-LTEs in a few years time then?
I dont care either way - whatever benefits the consumer the most really.
Are you serious?
So your saying the answer to sheety 3g coverage outside the M25 is to get rid of the 2G coverage and implement a whizzy new 4G?
Why not leap ahead and implement 38DD at least we'd all get a laugh from that.
Meanwhile HMG screws more shedloads of cash from the operators who in turn give us rubbish service for shed loads of dosh.
All I want is for my 3G or even GPRS to work only 20 miles from Colchester not the flippin outer hebrides FFS. And Colchester is a city BTW not a hamlet.
It's like the twilight zone out here and you muppets think instead of fixing the old one simply whacking out another poorly thought out and shoddily implemented bit of shiny kit is the answer? Good grief.
Fantastic article from Bill as usual, puts the whole history of mobile data in perspective very clearly and understandably without skimping on the technical info. Great read.
In the process of trying to make OFDM more palatable to a non-comms audience, you seem to have missed the real advantage of the technique. The real beauty of OFDM is the flexibility given by so many orthogonal low-bandwidth sub-carriers: you can avoid the problem of frequency-selective fading for one user (mitigating their bad channel) by moving their allocation in the spectrum. This leads to much more efficient usage of spectrum, and better channels for all. It is excellent at adapting to nasty channel conditions in slowly-changing channels*.
I'm not really sure what you're getting at with regards to "solving" the problem of "timing". If you're referring to inter-symbol interference, which is indeed a concern, then OFDM doesn't directly solve this by "putting [chunks] in different frequencies" - but what it does do is ensure that the symbol rate on each individual component frequency is low enough that ISI can be mitigated easily. But if you're referring to interleaving, which does indeed spread "chunks" of data amongst frequencies, then that helps solve a different problem: the issue of losing bursts of data in frequency-specific fades.
On the down side, OFDM really isn't good in channels with a great deal of doppler shift (since it breaks the orthogonality between the sub-carriers) - making it non-ideal for use in fast-moving vehicles, for instance, without specific counter-measures.
Seems like a decent analysis of the technology that connects the handset to the basestation.
1) Not a hint of the "killer app(s)" that will make people actually pay for this service. Whatever end users pay in the UK likely has to include a premium towards paying off the ridiculous prices the service providers paid in the 3G licence auctions.
2) Not a hint of how the data gets from the basestation to the service provider and thence the big wide world - what's the point in a 10Mbit link between basestation and handset when in much of the country the basestation connects to the service provider via nothing more sophistacted than (say) a DSL line with 512k upstream and (say) 2Mbit downstream. That'll work well when three people in that cell try to use (say) iPlayer simultaneously won't it.
What are the answers to those questions?
Gotta go, my first ever mobile broadband dongle arrived today and I need to play<<<<configure it (last week my landline died and I lost voice and data comms at home). Not compeletely new to this mobile data game though, I think I remember HSCSD on a NK6150.
Reading it was like being beaten with the softest of clue bats for 8 pages. Good work
All very well having the low-level hardware in place
And all credit to the RF engineers etc.,
But what's the point if the mobile operators then stick flaky http cacheing / image compression on top which frequently leads to web-page load-times measured in the tens of seconds (even with a basic 1.5Mbps "bulk data" throughput)... and corrupts ZIP downloads to boot?
(Yes, I'm a frustrated mobile-broadband user)
Oh I could say so much
about this issue and what ends some will go to protecting their investment. Its now a process whereby we're stuck with what the OFCOM quango wants us to have, via heavily lobbied polys'. The reality could be so much different.
And it wasn't named 802.11 "R" for nothing.....
@Chris Lewis - Finally someone who knows what he's talking about! Thx R
GT - pseudonym
that covers a lot of technical ground very well - despite the odd error. The area I would single out is the tradeoff between frequency and bandwidth: yes, lower frequencies propagate further but you have a lot less bandwidth to play with at 900 MHz than you do at 2 MHz.
Mobile data failed to take off largely down to the cost and the difficult of using it - we were all supposed to do everything on the phone. The cost might have been less of an apparent problem if we all still had only 56k modems at home but broadband signficantly changed both experience and expectations and then all of a sudden there was the idea of "free" wifi, subject to availability, ie. in a densely populated area oh, and video on the interweb became popular.
That said, 3G mobile is really wonderful in general as long as we accept that speeds will vary. For surfing (without video) and chat and e-mail I've found little reason to complain. And while I think we'll always hope for more the novelty has worn off a bit and some of at least have learned to wait.
EVDO and US 3G situation
@Charlie Clark, well, yes and no. There's certainly more spectrum at 2ghz to give out, but if you get say 20mhz of spectrum, that's 20mhz of spectrum if it's at 2.1ghz or at 600mhz, either way.
Good article. I should point out, in the US most 3G is EVDO Rev A (3.1mbits/sec down and 1.8mbits/sec up). Going from EVDO Rev 0 to Rev A was a software update at the sites so there's not much EVDO Rev 0 left. EVDO Rev B and EVDO Rev C, Qualcomm's got them ready, but no carriers have plans to deploy them -- Sprint's deploying Wimax, and most other CDMA+EVDO carriers here are planning to deploy LTE (NO, I repeat, NO legacy GSM -- just LTE using a pure IP backbone.) In cases where they need more capacity, they are just turning on additonal EVDO channels. UMTS here is really not worth mentioning -- AT&T has like 20% of their network upgraded to UMTS, T-Mobile has it only in a few markets; as a bonus AT&T's UMTS is reportedly quite flakey and slow. In contrast, Sprint has about ~60% of their network upgraded to EVDO, Verizon was nearly 100% pre-Alltel buyout, Alltel was about 60%, and even the regional CDMA carriers (US Cellular, Cellular South, Bluegrass Cellular, etc.) have extensive EVDO buildouts. It's normal here for these carriers to have "free" data roaming off each other too, so if I go to Kentucky for instance, I could roam off Bluegrass at full speeds.
I thirdly must LOL @ UK 3G coverage. As much as the US is generally slogged on about cell phone sophistication, etc. etc., in this case the tides have turned -- look at Verizon Wireless's 3G coverage map and see what I mean. I can go to my grandparents, a 950 mile trip, and have EVDO for about 940 of that 950 miles (about 8 miles are the 144kbps 1X data, and about 2 miles "no service").. It should be less than a year for Verizon to finish the EVDO upgrades (on the remaining non-upgraded Alltel coverage), but even as it is I should be able to go from here to the *west* coast (~2000 miles) with about 1800-1900 miles of EVDO coverage.. And (despite having an older EVDO Rev 0 card) I get over 500kbps almost all the time and 1mbit pretty regularly. (Forget loading web pages and e-mail, I can stream youtube and hulu pretty reliably.)
@Are you serious?
Colchester lacks what it takes to be an English city - a cathedral.
Still, with a population of ~100K (IIRC), it should be entitled to a decent mobile service, cathedral or not.
Handset or PC?
To me the main reason that data on 2G/3G etc. has not really taken off is that most content is not suitable for a phone handset.
When WAP first came out, most web content was too complex and there were attempts to produce WAP portals which would offer cut down content suitable for the handsets.
At that point most web content would have displayed well on my EEE PC (if it had existed then).
Now content is so rich and so loaded with fancy video and special effects that I struggle on my EEE PC and my older portable to view the website in the way that the cutting edge design intended.
Nice on my new 1440 * 900 Dell portable, though :-)
Not much good on a phone handset :-(
For me, mobile data is all about freeing portable PCs from fixed ADSL/cable connections.
You can get the rich content you have come to expect when you are away from your cosy nest.
Now if only the coverage, reliability, bandwidth etc. etc. wasn't totally crap. (Speaking as a Virgin customer).
If I want to talk to someone or send a text message I use my phone.
For getting serious data off t'Internet I use a PC.
Then again, I am from the keyboard generation and can't think through my thumbs.
"mobile data is all about ..."
That's the 64billion dollar question, isn't it.
Is it about seeing the same internet as at home (or in the office), when you're on the move (nb on the move, *not* just away from base but stationary at a WiFi hotspot).
Or is it about the mobile web, about some selection of significant websites recognising that there are going to be as many folk viewing them on "mobile internet devices" (phones, PDAs, maybe netbooks) as there are on PCs, and that their website designs should reflect that (eg no Flash).
It isn't about "m-commerce" yet, or about "location dependent services", and folks have been trying that for a decade or so. Mind you, Google has recently changed the market rules, as it sometimes does, with Google Maps for Mobile and the things you can do with that.
Yahoo nearly works on my S60 mobile in Opera Mini. BBC news has a "low graphics" version too. Google Maps for Mobile is fantastic, though without a GPS it sometimes gets confused (but that's probably not Google's fault).
But not everyone does so well. Obviously anyone designing in Flash has wasted their time even more than usual. Some sites that you would imagine might be of particular interest to those out and about are so full of big-screen rubbish and scripts and so on that they are useless on a small screen device, mobile or otherwise. Classic examples would include weather forecasts from the Met Office and traffic reports from the Highways Agency.
So, what exactly is the killer app for mobile broadband? How is it going to make money for the cellcos, so they can pay for all that extra bandwidth to all those new cells?
I know, we could do location-dependent downloaded-on-demand high-definition video ringtone subscription service. Yeah, that'd work. Where's the Dragon's Den number.
Well? You got any better ideas?
8390? shouldnt that be 8310?
Good article, but as this article seems to have a UK slant to it.. the 8310 should have been the phone that was mentioned as being the first with GPRS in the UK. The 8390 was not even the first phone in the US with GPRS, a motorola timeport was, if i remember correctly.
Jobs - because the iPhone wasnt a world first in any area.