Spectrum Crunch

[Zor]

Link

Just thought people should be aware of this.

Zor

[Singular Intellect]

Standard tunnel vision of the pessimistic doomsayers.

People seem to have no problem finding problems and even apparently professing understanding of issues like exponentially increasing bandwidth consumption and use.

What is hard to pound into the skulls of such people is the exponentally increasing technological progress and innovation that possesses the power to solve such problems.

From a mere five minutes of Googling and investigation:

Wireless breakthrough: one frequency, multiple signals
Scientists show off the future of Wi-Fi - smash through 3Gbps barrier
The Solution For The Bandwidth Shortage? Better Silicon Devices
U.S. Navy Rides the Terahertz Wave to Next-Gen Electronics
Graphene-based terahertz devices: The wave of the future

As per usual, people extrapolate current trends and systems into the future and do not take into account other variables throughout the larger picture that also change and provide solutions to such problems.

One set of problems get solved, new ones are presented and tackled. This is progress, same old same old.

[YT300000]

The funny thing is, we haven't even really had to bother with full-on bandwidth efficient design yet, since our focus on power efficiency and higher-order modulation schemes is giving us more than enough capacity to play with. And we're getting exponentially better at dealing with high frequencies - modern LTE systems run at 5 GHz, at ranges which would be impossible just a few years ago. Obviously if everyone tries to place a call at once we'll run out of capacity, but that's always been the case, and it's getting better.

In short, I'm not too worried about OTA communications suddenly becoming clogged.

[Blayne]

Sweet, another Extra Credits fan, you ever come onto the official Mumble channel or forums?

Standard tunnel vision of the pessimistic doomsayers.

People seem to have no problem finding problems and even apparently professing understanding of issues like exponentially increasing bandwidth consumption and use.

What is hard to pound into the skulls of such people is the exponentally increasing technological progress and innovation that possesses the power to solve such problems.

From a mere five minutes of Googling and investigation:

Wireless breakthrough: one frequency, multiple signals
Scientists show off the future of Wi-Fi - smash through 3Gbps barrier
The Solution For The Bandwidth Shortage? Better Silicon Devices
U.S. Navy Rides the Terahertz Wave to Next-Gen Electronics
Graphene-based terahertz devices: The wave of the future

As per usual, people extrapolate current trends and systems into the future and do not take into account other variables throughout the larger picture that also change and provide solutions to such problems.

One set of problems get solved, new ones are presented and tackled. This is progress, same old same old.

I could search and replace this post for bandwidth to oil and it sounds eerily similar to Peak Oil nay saying, the problem is not so much the ability to place calls but the ability to stream games live via cloud gaming which is an order of magnitude greater in usage than placing calls or surfing the internet.

Yes, technological innovations make it unlikely that cellphones will one day suddenly stop working if too many place calls at once, but the problem, the concern here will be the inability for a majority of consumers to say play Crysis over their tablet on the bus to work or school.

Extra Credits tends to orient its episodes to perennial problems as they apply to the future of gaming, obviously cell phone use as a whole is largely fine.

[Zaune]

... the exponentally increasing technological progress and innovation that possesses the power to solve such problems...

Is not something it's prudent to take for granted.

Look at it this way. Even if all the new technologies in the articles you've linked to actually work, they've got to become something more tangible than a patent and a press release before they're actually useful. A bird in the hand, and all that.

And even then, once they get to the commercialisation stage, it's still going to take a number of years and a considerable amount of money to fully implement them. The new systems will have to be backwards compatible with the old equipment, or at least operate in such a way that they don't interfere with running it in parallel, and demand for bandwidth on those legacy systems will not stay static in the interim period between the new transmitters going online and the last mobile device that can't receive the new signals being retired.

Basically, when you're looking at a problem you know you're going to have in five or ten years, you can't afford to bank on solving it with tools you think you'll have ready to go in time.

[aerius]

Look at it this way. Even if all the new technologies in the articles you've linked to actually work, they've got to become something more tangible than a patent and a press release before they're actually useful. A bird in the hand, and all that.

And even then, once they get to the commercialisation stage, it's still going to take a number of years and a considerable amount of money to fully implement them.

Bingo. If even 10% of the technology we see in the research labs panned out in real life we'd be living in a utopia world by now. Even a lot of the more promising stuff runs into various problems such as scaling, cost, lack of funding, or even something as stupid as a legal dispute over the patents that ends up getting the damn thing buried for the next 30 years.

Until you get it to work in the real world and prove that it can scale at a cost that's affordable, you might as well be talking about nuclear fusion and free energy.

BTW, maybe SI would like to cost out the price of this new technology, and how the telecom companies will finance its rollout given their current & projected balancesheets and revenues. Cause bottomline is it ain't getting built if you can't pay for it.

[The Jester]

As an RF Engineer presently performing network design for a major network rollout which is (surprise, surprise) affected by spectrum issues, I can say with some authority that this is a very real problem facing network operators. And while the RF spectrum is not a consumable resource in that you'll never run out of it by continuing to use it (unlike oil), there is a very real, and very hard limit as to how much information can be transmitted over a given band in the presence of noise. This limit is given by the Shannon-Hartley Theorem (see http://en.wikipedia.org/wiki/Shannon–Hartley_theorem) and is much like the speed of light limit in that all you can do is get closer to the limit given by the theorem, but you're never going to be able to exceed it regardless of your modulation or error coding. While technological advances, such as the move from GPRS to WCDMA to HSPA to LTE do help greatly to improve our spectral efficiency, we're still stuck with a hard limit given by the Shannon-Hartley theorem.

Now aside from the Shannon-Hartley, there are a number of practical issues that operators need to also contend with. For example, consider how spectrum licenses are divided up in most countries. What you'll typically see is that the lower frequencies tend to used by broadcast services such as radio and television whereas operators have had to push higher frequencies when they rollout a new technology. The problem with higher frequencies is that there is an inverse square law between frequency and free space propagation distance. So for any given technology, if you double the frequency, you halve the propagation distance for the same received power (why do you think the broadcaster operates at lower frequencies in the first place). In order to meet customer expectations in data throughput at the cell edge, you therefore need to have more cells in a given area than if you were operating at a lower frequency. In metro areas (with all their radio-blocking building clutter) this isn't so much of an issue since your radio waves weren't going to propogate very far in the first place, but you still have to face the practical issue of acquiring a site to actually place your equipment which can have all sorts of problems.

Consider: you need to find a willing landowner that will let you install equipment. You need to negotiate rent with said landowner, who may have an asking price higher than what is commercially feasible for the operator. If you need to construct a new tower or extend an existing one, then you need to deal with councils and have your building approved. During this process, people can end up objecting because they are scared of evil, invisible radiation that's going to fry their brain, give them all cancer and new-born babies birth-defects, or they just might not like the visual impact, since towers tend to be ugly.

Do you know of any coverage holes in your city? Ever wonder to yourself why doesn't the operator cover this part of the city when they cover another part which is less densely populated? Well, odds on they probably have considered adding coverage, they've probably hit some brick wall with the practicality of getting the site built. Anybody who lives in Sydney will know that there is no coverage by any operator in the city loop tunnel. Now you know why.

On top of all this, after a site is built the RF engineers still need to optimise it in relation to the rest of the network to ensure a good customer experience and smooth network operation. This, along with the site acquisition and construction process, takes time and costs money. So while you may be getting more bang for your buck with economies of scale with your equipment and construction, you still need to pay in terms of engineer time and rent which don't scale so easily (especially the rent part, especially rent in the inner city).

Aside from all of these issues faced by operators, there is also a problem for handset manufacturers. Remember the days of GSM, when your handset used to last around a week before needing a recharge? Now we're starting to get lucky if our handsets last a whole day before needing to be tethered to a wall by a power lead (so much for mobility). Now if you went and bought yourself a HTC Thunderbolt (or any other similarly capable LTE smartphone), you'll probably find yourself very impressed by two things: 1. the download speed is fantastic when operating on LTE, and 2. the rate at which your battery is consumed is equally impressive. You see, while we've developed increasingly better technology for utilising our available bandwidth, it seems that battery technology has been having a lot of trouble keeping up the pace. Higher frequency bands and higher throughputs have their costs along with increasingly sophisticated, multi-core handsets and their, large battery-sucking displays. So even if we assume that network operators are able to provide all the bandwidth a consumer could ever want, they're still going to see their user experience degrade, simply by the fact that they're going to have to recharge their phones more and more frequently.

As you can see, this is a very real problem for network operators. While I doubt we'll see radio communications grind a halt from network congestion, we will see slower adoption of higher bandwidth applications for handsets over time due to these constraints.

[madd0ct0r]

Jester - what % of the ultimate limit are we running at now (at a rough estimate?)

[Blayne]

I think the episode throws up some graphs that may answer it on review; additionally the notion that technology will postpone or otherwise solve the problem I don't think can be at all accurate if current estimates puts us at running out of bandwidth by 2014; 2020 might be reasonable for new technology to be deployed but less than two years?

[Irbis]

I could search and replace this post for bandwidth to oil and it sounds eerily similar to Peak Oil nay saying, the problem is not so much the ability to place calls but the ability to stream games live via cloud gaming which is an order of magnitude greater in usage than placing calls or surfing the internet.

More like predictions that by year 2000 every street in Paris will be buried under 5 meters of horse manure from all the transportation needs that would arise by then. Which, admittedly, didn't happen*.

*Ok, we traded one problem for another, worse in some respects, but still.

[YT300000]

As an RF Engineer presently performing network design for a major network rollout which is (surprise, surprise) affected by spectrum issues, I can say with some authority that this is a very real problem facing network operators. And while the RF spectrum is not a consumable resource in that you'll never run out of it by continuing to use it (unlike oil), there is a very real, and very hard limit as to how much information can be transmitted over a given band in the presence of noise. This limit is given by the Shannon-Hartley Theorem (see http://en.wikipedia.org/wiki/Shannon–Hartley_theorem) and is much like the speed of light limit in that all you can do is get closer to the limit given by the theorem, but you're never going to be able to exceed it regardless of your modulation or error coding. While technological advances, such as the move from GPRS to WCDMA to HSPA to LTE do help greatly to improve our spectral efficiency, we're still stuck with a hard limit given by the Shannon-Hartley theorem.

I'm just at the beginning of my grad studies in RF engineering, but the situation doesn't seem quite so dire to me - in fact, it reminds me of the "power wall" that microprocessors hit in 2004, when the limits of clock frequencies prompted industry to start building slower multi-core processors. Yes, we can only multiplex so far, and schemes like OFDMA will only buy us a few years, but we're constantly getting better and better at noise correction and handling higher frequencies. The RF industry has grown up and is now in the same arena as the semiconductor industry, wherein we need to create new developments as fast as we can to keep up. It's a real problem, but exactly the kind that makes us produce our very best innovations. At least, that's my exuberantly positive take from only 2 years of actual RF experience.

[The Jester]

Jester - what % of the ultimate limit are we running at now (at a rough estimate?)

Well, with LTE in lab conditions and about 20 dB of Signal-to-Noise, LTE can provide (ignoring overhead an error coding) 100Mbit/s over 20 MHz of bandwidth, which translates to 5 bps/Hz. For 20 dB, the theoretical max per channel is around 6 bps/Hz. So approximately 80%. However, there are some technical tricks to sidestepping this barrier.

Also keep in mind that the limit is based on signal-to-noise, so if you can decrease noise in the system, or increase signal strength (more power, better antenna gain, beam-forming, etc), you can push the boundary up.

I'm just at the beginning of my grad studies in RF engineering, but the situation doesn't seem quite so dire to me - in fact, it reminds me of the "power wall" that microprocessors hit in 2004, when the limits of clock frequencies prompted industry to start building slower multi-core processors. Yes, we can only multiplex so far, and schemes like OFDMA will only buy us a few years, but we're constantly getting better and better at noise correction and handling higher frequencies. The RF industry has grown up and is now in the same arena as the semiconductor industry, wherein we need to create new developments as fast as we can to keep up. It's a real problem, but exactly the kind that makes us produce our very best innovations. At least, that's my exuberantly positive take from only 2 years of actual RF experience.

Wait until you start working in mobile network design.

Noise correction (I'm assuming you mean error correction) will only take you so far, and there's still a limit on higher frequencies and the practical issues I describe above. (Site Acquisition is a constant fight for an RF Engineer.)

We're not completely without hope though, so far the best solution we are actually deploying (so physically exists in real life, not some white-paper) is something called MIMO (multiple-input and multiple-output) which works on the idea that if you have multiple antennas at the transmitter and receiver, then you can have a channel for each antenna pair over the same band, using some mathematical trickery to separate the channels since your transmissions are coordinated. This can increase your bit rate by a multiple of your transmitter-receiver antenna pairs (seemingly "breaking" the Shannon limit). Though note that the gains from this are only linear (while consumer bandwidth demand is rising exponentially) and there is only so many antennas you can actually physically attach to a device.

I also would like to emphasise again that this isn't some far off problem that may affect us in the future. This is something that I have to deal with when I go into the office tomorrow. It is already affecting us now. It's just that as a consumer, you're not fully aware of what the actual symptoms of the problem look like, or you simply may not be affected in your day to day routine.

[YT300000]

Wait until you start working in mobile network design.

Haha, fair enough.

Noise correction (I'm assuming you mean error correction) will only take you so far, and there's still a limit on higher frequencies and the practical issues I describe above. (Site Acquisition is a constant fight for an RF Engineer.)

We're not completely without hope though, so far the best solution we are actually deploying (so physically exists in real life, not some white-paper) is something called MIMO (multiple-input and multiple-output) which works on the idea that if you have multiple antennas at the transmitter and receiver, then you can have a channel for each antenna pair over the same band, using some mathematical trickery to separate the channels since your transmissions are coordinated. This can increase your bit rate by a multiple of your transmitter-receiver antenna pairs (seemingly "breaking" the Shannon limit). Though note that the gains from this are only linear (while consumer bandwidth demand is rising exponentially) and there is only so many antennas you can actually physically attach to a device.

I also would like to emphasise again that this isn't some far off problem that may affect us in the future. This is something that I have to deal with when I go into the office tomorrow. It is already affecting us now. It's just that as a consumer, you're not fully aware of what the actual symptoms of the problem look like, or you simply may not be affected in your day to day routine.

I hadn't really considered difficulty in site acquisition, to be honest. Actually, I just started learning about macro and microdiversity last semester, and might get involved with it more in the coming years as my major focus is antenna design (the software stuff is still basically magic to me). I guess my attitude is that as long as the consumer doesn't realize that something's going on unless they're told, then we've done our jobs - it remains our problem, and not theirs.

[tim31]

Anyone who has renewed a contract with Vodafone AU will watch the video in the OP and say 'Yeah, I can live with that.'