Spectrum Fragmentation Means Pricier Mobile Networking

alphadogg writes "The plethora of spectrum bands used for LTE will result in more expensive devices, and also make the ability to roam globally using the technology less likely, according to industry organization GSM Association's research arm. Wireless Intelligence predicts there will be 38 different spectrum frequency combinations used in LTE deployments by 2015, thanks to ongoing spectrum auctions, license renewals and reallocation initiatives across a wide range of frequency bands. The number of combinations means economies of scale won't be as good and prices won't come down as much as they could if fewer spectrum bands were used as volumes increase, Joss Gillet, a senior analyst for Wireless Intelligence."

Comments

Software radios

[koan]

http://en.wikipedia.org/wiki/Software-defined_radio

Re:Software radios

[fsckmnky]

I was just about to mention SDR and Fractal Antennas.

That said, anything "new" will be "more expensive" in the near term. As soon as unit qtys ramp up and it becomes "the norm" it should be no more expensive than any device made now.

Re:Software radios

[stox]

Software works well for the back-end of the radio, ie. detector. The front-end and antenna are another story.

Re:Software radios

[fsckmnky]

And how is this any different than the people who camped outside for days waiting for the first iPhone ?

Early adopters pay more. It has always been this way.

Re:Software radios

[SuricouRaven]

Hear that slurping noise from your new phone? That's the sound of a battery being sucked dry.

Re:Software radios

[Grieviant]

So SDR is going to drive prices down? Not if we're to judge anything from the price of the development kits and software. It sounds like the end-all in theory, but realize that these devices generally work from (a lower) IF frequency down to baseband rather than being completely digital from RF all the way down. You still need an analog front-end in the receiver, and you get a bulkier, more power hungry, more expensive piece of hardware. The need for future-proof, reprogrammable radios is questionable when the hardware is obsolete in a matter of a few years.

Re:Software radios

[icebike]

Software works well for the back-end of the radio, ie. detector. The front-end and antenna are another story.

But antenna's of approximately the right length are almost as good as specifically tuned antennas, and the technology for dealing with multi-wavelength radios is growing by leaps and bounds. See this summary of Fractal Antennas. (full article is paid), as well as this article

Developed over the last 20 years, fractal antennas have proven to be a fundamentally important breakthrough in antenna technology. This technology has allowed for antennas that are more powerful, versatile and compact. Because a fractal antenna uses fractal geometry and builds a complex pattern from the repetition of a simple shape, the inherent qualities of fractals enable the production of high-performance antennas that are typically 50% to 75% smaller than traditional ones. Because antenna performance is attained through the geometry of the conductor, rather than with the accumulation of separate components or separate elements that inevitably increase complexity and potential points of failure, fractal antennas offer better reliability and lower cost than traditional antennas.

.

So it would seem, that these antennas are destined to simply be "Printed" onto a substrate, perhaps the back cover of the phone, and segments enabled as needed. One antenna for all bands, just by using different segments to create the best pattern. This is bound to become dirt cheap to make.

When combined with a software defined radio, rather than the discrete band models we are used to, the flexibility to produce a true world phone is possibly closer than previously thought. As soon as the designers stop chasing multi-discrete-band radios, and just plan for a world of hundreds of band segments, the Software Radio will drive the unit lower than what we have today.

Re:Software radios

[fsckmnky]

If virtually everything could be done with software on a power-sipping cpu

This is being done already. But instead of thinking CPU, as in desktop/laptop computer, think CPU, as in custom reconfigurable DSP/FPGA combination specifically designed for implementing radio.

Re:Software radios

[Anonymous Coward]

I'M AN ENGINEER DESIGNING CELLULAR HARDWARE.

Please let me re-iterate what the article says: "The plethora of spectrum bands used for LTE (Long-Term Evolution) will result in [***]more expensive devices[***], and also make the ability to roam globally using the technology less likely."

The issue is not the modulation / coding, which is already performed via software and unconverted (ie software defined radio). The issue is in the RF front-end, which is highly specific to its frequency of operation. This includes FILTERS that are tuned to precise frequencies of operation, POWER AMPLIFIERS that are tuned to a specific frequency (some have slight control over their frequency via IO pins, with trade-offs), and LOW-NOISE AMPLIFIERS to name a few. Anytime you try to make this hardware work over a wider band or become reconfigurable, there are serious trade-offs in efficiency, cost, and overall performance. Cell phone antennas are not the issue.

Software defined radios generally don't work in the GHz range. They produce signals in the MHz range and up-converted into the GHz range, but thank you Slashdot for trying to teach me how to design a phone! This software-defined fully-reconfigurable world phone is a brilliant idea! I'll let my boss know our next design will be the size of a spectrum analyzer, weigh 10 kg and cost $10,000. And you won't be able to talk and surf because the software defined radio won't have the bandwidth for both the LTE signals and the WiFi signals. Again, THANK YOU SLASHDOT FOR TEACHING ME HOW TO DESIGN A PHONE!

To add to the cost, forcing cell phone designers to re-design pieces of the phone for different regions will add to the price even more. There's a push to lower the cost by creating one design that can operate in any region (re-designs / modifications are expensive).

Re:Software radios

[rev0lt]

It seems that the parent forgot that there's an actually transmitting unit (and all the power required), and that fixed-frequency radios are actually dirt-cheap since the '20s. I baffled that we have tri/quad-frequency phones today, without being an expert. But you got mod down because you're an ass. I'm an ass too (but not speciallized on cellphone design, so no cynism there), and if I could I would mod you up. Not because you're an ass, but because the parent is so full of wrong it hurts, and you're actually right.

Re:Software radios

[RobbieThe1st]

Why not just put the radio on a card, perhaps like a SIM card? Then, you can swap in a card/module for the band/area you want to use.
And any company wanting a new frequency would simply need to design another card to a certain API/interconnection standard.

Re:Software radios

[LeDopore]

Not necessarily. This lab from Cornell isn't going to patent their SDR, and it's a pretty sweet implementation: noise performance almost as good and chip area only slightly larger than a single-band ASIC:

http://molnargroup.ece.cornell.edu/research.html

Full disclosure: I worked with these guys, and the radio group is pure genius.

Re:Software radios

[fsckmnky]

Are you saying patent royalties are the only expense when it comes to new technology ?

Re:Software radios

[blindseer]

If you must ask why the answer is usually money. Let's assume a company thinks this is a good idea and investigates the possibility. First thing they'd have to do is decide on the specifications on the card. How big would this card have to be? How much power will it consume, and therefore how much heat will it produce? What frequencies will it operate?

Just looking at the frequency issue the RF connection could be a problem. Materials can act very differently at 800 MHz than it does at 2.3 GHz. A mismatch can mean problems with power consumed, frequency stability, out of band noise, and perhaps other things I missed. One could place the antenna on the card to avoid much of these issues but then it's no longer a card that is inserted in the phone but now the back half of the phone.

Just thinking of the logistics the phone designer would have to have one part for the UI half of the phone and a part for each frequency of operation for the RF half of the phone. That means tracking N+1 parts instead of N parts when N equals the frequencies of operation. Having talked to engineers that have to track parts they do everything they can to keep part counts down. Just adding another different kind of machine screw to the design can give them fits.

I suspect that there is already a high level of reuse in the designs on these phones. One chip drives the OS on a number of similar phones, one chip drives the RF on a number of similar phones, one chip drives the display on a number of similar phones. The phone designer can pick and choose from these already existing chips to build a new phone and choosing to design a new chip when there is a desire to add some new feature. To make this common connection standard idea work a phone company would have to convince the chip manufacturers to design a chip to their liking, or the cell phone manufacturers would have to design a board for these chips to sit on to do the translation for them. When dealing with GHz frequencies, and the constraints of a hand held device, this becomes difficult and therefore expensive.

People's tastes in cell phones shift rapidly, and therefore the designs must come to market quickly. Taking the time to come up with some connection standard, have it tested, and have it future proofed against changes in FCC rules, consumer tastes, and so on, while still allowing for a profit margin sounds quite difficult to me.

Re:Software radios

[Maow]

I'M AN ENGINEER DESIGNING CELLULAR HARDWARE.

OKAY. IF YOU CAN MAKE, er, ahem, if you make valid points, we'll believe you, don't have to shout.

thank you Slashdot for trying to teach me how to design a phone!

THANK YOU SLASHDOT FOR TEACHING ME HOW TO DESIGN A PHONE!

Apparently you're also a dickhead. Who knows how to design a phone. But a dickhead none-the-less.

Re:Software radios

[icebike]

More likely just a really good bullshitter with a few cut and paste paragraphs.

Anytime anyone wants to bet against technology bringing down prices while increasing performance and reducing size and energy consumption you can be pretty well assured they haven't passed their 20th birthday.

Re:Software radios

[LeDopore]

I grant that it's always possible to do at least a little better when you know the frequencies you'll use at design time than if you have to have a software defined radio.

That said, Caroline Andrews of the lab I worked in (http://molnargroup.ece.cornell.edu/research.html) developed a software-defined radio using a passive mixer that works in the GHz range with only 1 dB worse noise for the same power as you would use with a fixed radio. She also showed how a software defined radio can do impedance matching through the passive mixer. Amazing, possibly game-changing stuff; if you really are a cellphone radio engineer, don't assume that software-defined radios will always suck hard enough not to be a threat.

By the way, on behalf of Slashdot, YOU'RE WELCOME FOR TEACHING YOU HOW TO DESIGN A PHONE!

Re:Software radios

[LeDopore]

Are you saying patent royalties are the only expense when it comes to new technology ?

No, but developing one SDR that works for all phones in 50 countries without worrying about patent infringement certainly sounds like a good way of getting a high ROI when investing in a new technology. Of course there's a lot of work and expense, but I think SDRs are still a great way to go, especially since the best ones now (unlike even 2 years ago) perform almost as well as hardware-defined radios.

Lol

[ExtremeSupreme]

You guys in the U.S. don't know how lucky you have it. In Canada we get ass-reamed. Even Virgin Mobile Canada charges $35/month just for 500 megabytes of data.. my understanding is that in the U.S. that would get you heaps of data and a few hundred minutes. You might have it bad, but here in Canada we have it far, far worse!

Re:Lol

[sgt scrub]

Um yeah, yeah. We have it good here in the U.S. Gigabit wireless unlimited time, unlimited bandwidth. We also have some nice property opportunities. In fact, I have a nice water front property with its own private bridge available for a small down payment of, lets say, $2,000 U.S.D. Please send the check to.... You have typed more than 40 characters. Your usage limit has been exceeded. Your 14.4k modem connection has been terminated. -- AT&T

Re:Lol

[rtaylor]

If you're in a major Canadian city then you have another choice until Christmas at least.

http://www.windmobile.ca/en/Pages/Holiday-Miracle.aspx

Re:Lol

[Osgeld]

hm, in the US that I live in 35 bucks a month gets you no data limited voice and unlimited sms, or at most 30 megs, so dont be a hoser

Re:Lol

[Briareos]

I'm paying 9.90 EUR a month for 1000 minutes of voice, 1000 SMS and 1GB of 3G data, with 0.04 EUR for every extra minute or SMS (not that I'd ever reach those) and 4 EUR for every following GB...

Are you Americans still living in the dark mobile ages or something?

Re:Lol

[Maow]

You guys in the U.S. don't know how lucky you have it. In Canada we get ass-reamed. Even Virgin Mobile Canada charges $35/month just for 500 megabytes of data..

Won't dispute that Rogers, Bell, & Telus are rip-offs, and their Fido, Solo, & Koodo sub-brands are a joke, and Chatr is criminal in its anti-competitiveness (driving out Wind & Mobilicity), but with Wind I spend $40/m and get unlimited talk & text, Canada-wide, US-wide (not sure about Hawaii), unlimited global SMS, unlimited MMS, and unlimited data.

You cannot get that in USA as far as I know.

my understanding is that in the U.S. that would get you heaps of data and a few hundred minutes. You might have it bad, but here in Canada we have it far, far worse!

You may want to re-evaluate your understanding. Unless you have a hard-on for all things American (hey, there's a fair bit of that up here, so whatever turns your crank...), their mobile market is really not much better than ours at all. Maybe worse since we have Wind & Mobilicity.

We're absolutely indistinguishable if you look at European / Asian mobile markets and then compare back to Canada / US.

Ownership of Spectrum is simply wrong..

[yossie]

Radio Spectrum starts out as public property - selling it, a limited resource, with unknown but extensive economic effects in the future, is a bad idea. I strongly believe we should nationalize radio spectrum and lease it out, for limited periods (a dozen years? two dozen, tops.) to companies to use as part of their product.. Given how fast technology moves, and how useful (in unknowable ways) Radio Spectrum will be in the future, we are selling away our birthright..

Re:Ownership of Spectrum is simply wrong..

[fsckmnky]

Its not "sold" .. its licensed. They are also working to reclaim, and re-license unused parts of the spectrum.

Placing an arbitrary cap on the length of time a company can recover their expense, will have those unintended consequences you claim you don't want.

Re:Ownership of Spectrum is simply wrong..

[Anonymous Coward]

You have it backwards. Our government is criminalizing use of something that by definition cannot be owned. There is no excuse for violently restricting people from access to creating products that use certain ranges. Naturally we want cooperation of usage, but that is not what is going on right now. Nationalization (that is, even more strict violence against those who are doing nothing wrong) does not solve this problem.

From an appeal to effect, the case can also be made for letting society work out how best to use this range of frequencies. Economies of scale through fewer businesses in the market must be compared to efficiencies of competition as well as a number of other factors that determine how many competitors are optimal in a given market, in the eyes of consumers. The usage arrangements must be determined from the bottom up, not top down, if we are to achieve best possible results.

Giving state monopoly to a privileged few is not in the interest of consumers. We have enough examples that support this simple truth already: http://mises.org/journals/rae/pdf/RAE9_2_3.pdf

Re:Ownership of Spectrum is simply wrong..

[Solandri]

Placing an arbitrary cap on the length of time a company can recover their expense, will have those unintended consequences you claim you don't want.

I suspect you've never run a business, or at least not handled the accounting side of one. Placing an arbitrary cap on the length of time a company can recover their expenses is exactly what you want.

The ROI from owning (licensing) spectrum is a rate: e.g. dollars per year.
The cost of the spectrum is an amount: e.g. dollars.

The only way to reconcile these two is to either:

Limit the amount of time they can use the spectrum. Then:
ROI = (dollars / year) * (years) = dollars
Cost = dollars

or

Make the licensing cost a recurring annual fee, not a one time payment. Then:
ROI = dollars / year
Cost = dollars / year

Only when the units for cost and return are consistent can you make an analytical fiscal decision. Even purchases with a one-time fee, like a car, are turned into rates in accounting. You amortize the car's cost over the number of years you expect the car to remain in service. So if the company buys a car with a loan whose total payments work out to $35k, and you expect to use the car for 7 years, then the cost of the car is $35k / 7 years = $5k per year.

Any cap you place won't be arbitrary. It will taken into account in the bidding process. If a company thinks they can make $1 million/yr from the spectrum, and you place an arbitrary cap of 5 years, then they will not bid more than (assuming 10% profit margin) $4.5 million minus interest. If your cap is 10 years, then they will not bid more than $9 million minus interest.

As a fiscal conservative who has run a business and done the accounting for it, our government's insistence on auctioning spectrum in perpetuity for a one-time fee has always baffled me. It's like saying if you pay me $1000 one time, I will clean your bathroom once a week forever. It makes no business sense because it's impossible to tell if I'm getting a good deal (maybe you'll die next week) or a bad deal (maybe you'll live to be 120).

Re:Ownership of Spectrum is simply wrong..

[ironjaw33]

I suspect you've never run a business, or at least not handled the accounting side of one. Placing an arbitrary cap on the length of time a company can recover their expenses is exactly what you want.

I suspect you've never managed an economy. If you place a cap on the amount of time a company can recover its spectrum expense, you limit the scope and scale of potential services for that block of spectrum. Furthermore, as the government ( the public ) is the one holding the auction, placing a time limit also caps the amount of money that can be raised for the treasury today. If you want to experience the joyous wonders of a planned economy, move to North Korea. Check your vast expanse of business acumen at the door.

This sounds like the same argument for extending patents and copyrights indefinitely. Except in this case the medium is in the public domain from the start and there's only a finite quantity of it. A time cap of _appropriate_ length will incentivize the licensees to maximize the use of the medium rather than let it go to waste.

Re:Ownership of Spectrum is simply wrong..

[fonos]

Accounting is great, however it uses a few tricks that allow costs contributing to a product to be expensed when revenue is made. The most important thing is cash flows, because cash actually adds value to the firm.

The problem with your analysis is that it doesn't take into account the time value of money. A dollar today is worth more than a dollar in a year, simply because putting a dollar in the bank gives you more than a dollar in year. If the present value of the expected future benefits are greater than the present value of expected costs, as long as your cost of capital is correct the project WILL add value to the firm, regardless of how long it takes. this is called NPV analysis.

The payback period method on the other hand is not based on Economic theory, it doesnt take into account cash flows after the payback period, favors small projects and discriminates against large ones. There's no Economic method to calculate a proper payback period, and so the payback periods are arbitrary.

Many businesses have disregarded great projects that would have added value to the firm but simply had the wrong accept/reject criterion. Accounting methods are not quite appropriate here, since the main point of accounting is to match costs with revenues, and indeed the financial formulas to find a stock's price involve taking the accounting financial statements and working backwards to find the real cash flows again.

About perpetuity, imagine that buying the spectrum was equal to receiving a set number of dollarsfrom the government every year, let's say $1000/year forever. Assume a 6% cost of capital for the company. Now we can easily find out how much this is worth today. 1000/.06 = $16,666.67. (This is a limit simplified down, 1000/(1.06) + 1000/(1.06)^2 + 1000/(1.06)^3 + ...)

That would be the cost you need to pay today to get a perpetual annuity of $1000, kind of like buying rights to the spectrum.

Are you still baffled? It's really just Economics backed with maths.

Re:Ownership of Spectrum is simply wrong..

[Anonymous Coward]

No. Markets are proven to be the best way to deal with scarcity (no matter scarce the resource). Auction all spectrum to private owners (for good), and get rid of the FCC. I don't want a government bureaucracy determining the future of something best left to industry leaders who actually stand to lose something if they screw up.

Re:Ownership of Spectrum is simply wrong..

[Bitmanhome]

As a fiscal conservative who has run a business and done the accounting for it, our government's insistence on auctioning spectrum in perpetuity for a one-time fee has always baffled me.

It's not baffling at all, it's simple corruption. It's abuse of a public resource for private gain.

Re:Ownership of Spectrum is simply wrong..

[chromaexcursion]

You have no clue! Ownership of spectrum is absolutely required. It's called economics. No company will invest in infrastructure without it.

Re:Ownership of Spectrum is simply wrong..

[fsckmnky]

It's abuse of a public resource for private gain.

When the government licenses the spectrum to a public company, financed by public investors, in exchange for public funds, so that the public company, can offer services to its public customers, exactly how much more friggin public can you make the process of utilizing a public resource ?

Did you go to public school or something ?

close...

[msauve]

but not quite.

"The number of combinations means economies of scale won't be as good"

Prices may be more than if there were fewer frequencies, but I'd expect frequency agile chipsets, able to handle the full range of LTE frequencies, to be manufactured. So it's not a case of economy of scale, but the COG being a bit higher because of more capability.

Re:close...

[mcelrath]

I was just wondering this the other day...it seems to me that for a handset manufacturer it would make sense to put all of CDMA/TDMA/GSM/LTE/HSPA+ etc onto one chip, and define the frequencies and protocol by some BIOS settings. That way the same phone could be sold to every mobile carrier. I would think it should also be possible to include many antennae or fractal antennae.

Is this already going on? Or are handset manufacturers really putting different chips in the same handset destined for different carriers?

Re:close...

[ColdWetDog]

I was just wondering this the other day...it seems to me that for a handset manufacturer it would make sense to put all of CDMA/TDMA/GSM/LTE/HSPA+ etc onto one chip, and define the frequencies and protocol by some BIOS settings. That way the same phone could be sold to every mobile carrier. I would think it should also be possible to include many antennae or fractal antennae.

Is this already going on? Or are handset manufacturers really putting different chips in the same handset destined for different carriers?

That's hard to do with the frequencies being significantly different. LTE spectrum ranges from 850 mHz to 2500 mHz - a wide spread to handle in a tiny radio. It's possible to do but then you have the annoying engineering tradeoffs of size, battery efficiency and cost.

Re:close...

[amram9999]

Manufacturers would do that if they could, but it's not technically feasible given space, cost, and electrical constraints. That's the whole point of the article. Your antenna needs to be tuned for every frequency band that you support, so there is a finite number that you can support well. Also, the power amplifier must support each frequency band. Also, the RF receiver must support each band and each protocol in hardware. It's not trivial at all to add a new frequency and/or protocol.

Re:close...

[msauve]

It's not particularly hard to do on a chipset, using DDS technology. What is necessary, are external filters for the specific frequencies/bands, but stuffing different filters/antennas on essentially the same board doesn't really hurt economies of scale, when each band may represent millions of handsets. There's not much economy of scale gained between building 1 million of something, and building 2 million of the same thing, at least when you're talking about $50+ things (i.e. at that scale, you're saving pennies, not dollars).

Re:close...

[whoever57]

What is necessary, are external filters for the specific frequencies/bands, but stuffing different filters/antennas on essentially the same board doesn't really hurt economies of scale,

That's fine as long as you plan to use your phone only in one country and one operator.

Re:close...

[Tacvek]

Your point is correct, but your math looks a little off.

850 mHz= 850 milliHertz = .85 Hz = a little under once per second.

2500 mHz = 2500 milliHertz = 2.5 Hz = two and a half times per second.

Still very low bitrates, especially since we are talking about the carrier frequency, so the data rate will be much lower.

Re:close...

[the_other_chewey]

Your point is correct, but your math looks a little off.

850 mHz= 850 milliHertz = .85 Hz = a little under once per second.

2500 mHz = 2500 milliHertz = 2.5 Hz = two and a half times per second.

Ah, crap, yes, you're absolutely right.

Somehow, I managed to base my numbers on periods of "every $mHz seconds",
which is total bollocks of course. I don't really have an excuse for that right now...

Wait

[should_be_linear]

why is problem to negotiate spectrum availability when ID of provider changes? This non-issue was solved decades ago.

Re:Wait

[russotto]

Because this article is a thinly-veiled plea to give all the spectrum to Verizon.

Re:Wait

[stephanruby]

No, I think the opposite is true.

Verizon may have become a member of the GSM Association just this last year, but it's still using CDMA, plus its own brand of 4G, so in effect the GSMA is probably criticizing Verizon. After all, the entire World has already standardized on GSM, even Mainland China has, but there still remains one or two hold outs here in the US, mainly Verizon and Sprint. Those two don't want to play in the same sandbox with others (although, they have to when their business customers ask for World phones).

Refuse

[ebonum]

Can the handset makers put their foot down? Can they simply say: we only support this list of 5 major frequencies that are used worldwide? Phones would be smaller if they didn't have to be 5-band (or more).

Seriously, all these frequencies is stupid. People need phones that work world wide.

If I was Samsung and Verizon wants a phone for their LTE network, I would tell them. 1) you must promise now to buy x ( 1 million? ) at an inflated price. 2) You must pay Y million ( 50 million? ) today to cover our development costs to build a phone that works on only your stupid network. 3) We have the right to make it a one band phone. If your network coverage is bad, it will not be able to switch over to another network.

Re:Refuse

[mollymoo]

Are engineers going to work out a way to get a couple of orders of magnitude more data in the frequencies they've already got? No, so we need more frequency bands to significantly increase capacity. Are Nokia are going to get Switzerland to change the frequencies their emergency services use so they can save $0.05 on the cost of making a phone? No, so we can't use the same frequency bands everywhere in the world because there just aren't that many big chunks of useful spectrum lying unused.

Re:Refuse

[gl4ss]

the phones wouldn't be any smaller than what they are now, tbh. the cell chip is pretty tiny already.

battery and display make the most of the phones size and thickness nowadays - the days of trying to make the phone smaller and smaller went away with the zippo nokia 12 years ago.

Re:Refuse

[demiurg]

Whatever major handset manufacturer will refuse to do, Huawei, ZTE and other Chinese vendors will do overnight and take over the market.

Re:LTE is rubbish.

[Overzeetop]

Um, excess bandwidth charges?

It's all I can figure is the endgame here. I mean, I just sat in a presentation about the future of LTE going up to 1Gbps. On my plan, that means I could hit my limit and get charged another $15 in just a little under 2 seconds.

Re:LTE is rubbish.

[danbob999]

The latency is too high in UMTS. VoIP isn't reliable enough, yet.