Future Wireless Bandwidth

[SkepticJ]

What is the limit to wireless data transfer rate via radio waves? Forget the technolgy, it's unknowable. Simply what can the electromagnetic radiation physically carry?

Is it conceivable that in the year 2150, cell tower networks, blimps, satellites, or whatever they're using then, could support 1 TBps rates to individual users?

This may seem like an insane rate, but our smartphones that drink in MBps would be inconceivable to a wireless telegraph operator from a century ago. Who knows what they'll use it for? Real-world-fidelity VR, or something we haven't even thought of yet.

[LookingSkyward]

Data rate is dependant on bandwith, roughly, so if we can run a really high frequency with a wide bandwidth, the technology is the limiting factor. I vote for using light I vote for laser to blimp uplink phones.

I'm sure someone with more knowledge can provide a more detailed answer, but the original ethernet was over radio.

[Swift]

On the flip side, what does one need? I can't even imagine of what I would need transfered to my hand at 1 TBps rates.

[MikeMcc]

I did see something recently on science daily about a new modulation method that would improve the data capacity significantly. Cant find the link at the moment unfortunately.

[Ara Pacis]

I did see something recently on science daily about a new modulation method that would improve the data capacity significantly. Cant find the link at the moment unfortunately.

Would that be helical waves? http://news.discovery.com/tech/radio...st-120302.html

[SkepticJ]

I vote for laser to blimp uplink phones.

The big problem with that is you would need line-of-sight. Radio can pass through opaque materials.

[SkepticJ]

On the flip side, what does one need? I can't even imagine of what I would need transfered to my hand at 1 TBps rates.

Our smartphones that drink in MBps would be inconceivable to a wireless telegraph operator from a century ago.

Who says it would be transferred to your hand? I highly doubt smartphones are the end, any more than wireless telegraph, optical telegraph, or smoke signals were the end.

I mentioned one possible use. Transmitting an artificial construct that has the fidelity of reality would gobble it up. Especially considering that current humans only perceive small slices of the universe as it truly is. We can see a small part of the EM spectrum at a pathetic resolution, and we have four other gross senses. It's conceivable that VR universes could be even more complex than the real world. For example, one could be free to move (and see, etc.) in more than three spatial dimensions. Different "physical laws" could apply to one virtual entity, but not another. The possibilities are really endless.

My basic point is that we can't know what they would need. Who can see the future?

[danscope]

The question is " Need " . You can buy a formula 1 Ferrari , but the speed limit is 65 most places. Duh....You're sitting on a
settles plateau. With full motion video point to point, you don't really "Need" the performance. Who wants to pay for it.

[Noclevername]

The same people who pay for HD and plasma screens today?

[Solfe]

I don't think one TB a second in 2150 would be too wildly out of line. My 1999 iBook has a 20ish GB hard drive and less than 500 MB of ram. Today, you can have an ibook with 8 GB of ram (16+X more) and 1 TB drive (50x increase) without completely breaking the bank. My 1999 iBook as it is now would have broken the bank back in 1999.

[Tuckerfan]

Bill Joy of some software company (Oracle?) did a bunch of experiments with wireless communication back in the late '90s (Wired magazine did an interview with him about it) and according to him, the upper limit is dictated by processor speed of the receiving device. Once you exceed the speed of the processor, there's simply no point in trying to push any more data.

As for why you'd "want" such massively high transmission speeds, Ultra High Definition TV is expected to hit the shelves in 2020 or so. An hour long video will be several terabytes worth of data, more if frame rates higher than 120 fps catch on. You can expect the camera in your cellphone to be HDTV quality (or better) by 2020, so super high speed wireless is going to be an absolute necessity by then. (Not to mention all the other data hogging applications which will be on our phones and other devices by then.)

[caveman1917]

On the flip side, what does one need?

Well, 640K ought to be enough for anybody <wink>

[Ara Pacis]

Well, 640K ought to be enough for anybody <wink>

There's only going to be a worldwide market of 4 or 5.

[Squink]

On the flip side, what does one need?

Rough calc for binocular artificial reality at the full resolution of human vision:
Typical human eye resolution, according to Plait, is about 1 arcminute.
A sphere measures 148,510,660 square arcminutes. The Human eye has a field of view of somewhat less than 180°; so 148,000,000 pixels should suffice to produce undetectably pixellated static display for both eyes.
32 to 64 bit pixel depth should satisfy color requirements.
60 frames per second should eliminate frame change jerkiness for most people.
148,000,000 X 64 X 60 = 588 Gigabits per second.
Add stereo sound and error correction to that, and you might get near a need for 1 Tb per second data transfer.
Remote processing of the visual field to, for example, edit all cats out of the user's reality, would easily require bandwidth in excess of a Tb/sec.

Still need a factor of 8 or so to get into the TB/sec range though. It's not obvious what could use that much more data, even multiplexing between multiple virtual realities would still require only Tb/sec, and tactile suits seem unlikely to ever require even that amount of i/o.

[Ara Pacis]

Rough calc for binocular artificial reality at the full resolution of human vision:
Typical human eye resolution, according to Plait, is about 1 arcminute.
A sphere measures 148,510,660 square arcminutes. The Human eye has a field of view of somewhat less than 180°; so 148,000,000 pixels should suffice to produce undetectably pixellated static display for both eyes.
32 to 64 bit pixel depth should satisfy color requirements.
60 frames per second should eliminate frame change jerkiness for most people.
148,000,000 X 64 X 60 = 588 Gigabits per second.
Add stereo sound and error correction to that, and you might get near a need for 1 TB per second data transfer.
Remote processing of the visual field to, for example, edit all cats out of the user's reality, would easily require bandwidth in excess of a Tb/sec.

According to this, the acuity is actually even better, at about .3, making about 576 megapixels at 120 degrees. To get 180, you'll need to increase the pixels. Of course, the brain fills in gaps all the time (such as the human blind spot), so it could be less. If you consider that a lot of detail is unavailable due to haze and fog and blurring from being out of focus or in peripheral vision, you can reduce that. And all this assumes no compression, which could be lossy and still high fidelity. And then you need to consider that the resolution only needs to be a certain number in the display system. A lot of processing can be performed locally to convert vector images and object generation into bitmaps for the display, further reducing the bandwidth requirements.

[HenrikOlsen]

Still need a factor of 8 or so to get into the TB/sec range though. It's not obvious what could use that much more data, even multiplexing between multiple virtual realities would still require only Tb/sec, and tactile suits seem unlikely to ever require even that amount of i/o.

On that other hand, at that time you might well be subsidizing the phone partially by having it's spare capacity be used for distributed computing, with apps being massively parallel things running on multiple phones and in that scenario bandwidth is a limiting factor again even at TB/s.

[SkepticJ]

Still need a factor of 8 or so to get into the TB/sec range though. It's not obvious what could use that much more data, even multiplexing between multiple virtual realities would still require only Tb/sec, and tactile suits seem unlikely to ever require even that amount of i/o.

You'd probably want at least several times the rate that is required under optimal conditions, that way if you're not getting a good signal, your VR isn't stuttering.

[Ara Pacis]

Add stereo sound...

Actually, it should be binaural for actual VR, but it would take the same bandwidth if calculated upstream. It might be processed locally if there is a way to synthesize it, but that might mean sending all the sounds separately, which would increase the bandwidth.

[SkepticJ]

Getting back to the OP, what determines what a given bandwidth of EM can carry?

Is it the number of photons per second, along with their hertz? Photons with a gigahertz wavelength can carry more than megahertz waves? Microwaves can carry more per-photon than radio, IR more than microwaves, visible light more than IR, UV more than visible, etc.? Or do I misunderstand?

[billslugg]

The absolute limit to the information carrying capacity of a system is limited by the bandwidth. There is no solution to that limit. It cannot be surpassed no matter what you do. Compression does not count. It is simply doing with less information.

In very rough terms, as I vaguely remember it from 40 years ago in EE??? "Communication Theory", the bit rate is about the same as the number of complete wave forms per second. At a THz you can transmit roughly 1Tb/s assuming that the bandwidth of the device extends from DC to 1THz.

[JCoyote]

At a THz you can transmit roughly 1Tb/s assuming that the bandwidth of the device extends from DC to 1THz.

This feels way, way off the mark. The limits are based on the number of discrete frequencies and the total detectable overall signal changes we can transmit along each and across all channels a given pair of devices can share. For the spectrum you mentioned it should be far higher than that.

However, local processing will likely remove the suggested need for such tremendous bandwidth. Look to current MMORPGs and FPS games... they do not transmit the picture, nor does the picture need to be transmitted; only the fundamental features need to be sent then the local processor and storage intelligently assemble it into a fluid reality. Finite geometry and distinguishing features take an awful lot less bandwidth to send.

[publiusr]

More on spiral waves

http://www.extremetech.com/extreme/1...ity-infinitely
http://www.grepscience.com/archives/...k-ever-created
http://www.space.com/16675-nasa-sign...-universe.html

[cjl]

In very rough terms, as I vaguely remember it from 40 years ago in EE??? "Communication Theory", the bit rate is about the same as the number of complete wave forms per second. At a THz you can transmit roughly 1Tb/s assuming that the bandwidth of the device extends from DC to 1THz.

I agree about this number seeming way low - even the modern (brand new) 802.11ac wireless standard involves transmiting 500-1000 Mbps or more over a 5GHz signal with 80-160MHz channel bandwidth. That alone is at least a factor of 10 better than your estimation, and it is achievable with current technology (you can actually go out and buy an 802.11ac router right now, though not very many devices support it).

[Squink]

Bandwidth is the key difference here. The calc is for AM, not FM.

[Solfe]

A tangent - Why don't wireless devices work on multiple channels/frequencies? I see many devices are able to use multiple channels/frequencies, but never at the same time.

[SkepticJ]

More on spiral waves

http://www.extremetech.com/extreme/1...ity-infinitely
http://www.grepscience.com/archives/...k-ever-created
http://www.space.com/16675-nasa-sign...-universe.html

Very interesting.

That about wraps this thread up. I mean where can one go from theoretically infinite?

[Tuckerfan]

A tangent - Why don't wireless devices work on multiple channels/frequencies? I see many devices are able to use multiple channels/frequencies, but never at the same time.

Interference issues. If they transmit on frequencies too close to one another the signal will be garbled. On portable devices there's also the issue of battery life. Radios can draw a lot of power and seeing how as everyone can't lift a phone that weighs more than three ounces, manufacturers are loathe to put bigger batteries into phones.

[billslugg]

I agree about this number seeming way low - even the modern (brand new) 802.11ac wireless standard involves transmiting 500-1000 Mbps or more over a 5GHz signal with 80-160MHz channel bandwidth. That alone is at least a factor of 10 better than your estimation, and it is achievable with current technology (you can actually go out and buy an 802.11ac router right now, though not very many devices support it).

Here is a quote from the Wiki page on Bandwidth.

"Hartley then combined the above quantification with Nyquist's observation that the number of independent pulses that could be put through a channel of bandwidth B hertz was 2B pulses per second, to arrive at his quantitative measure for achievable line rate.."

For a 1 THz signal, the baud rate would be 2THz. I don't know how to explain the 1GHz rate on a 160 MHz bandwidth. My prediction would be that it would only handle .32 GHz, not one GHz.

Perhaps the device is using some form of compression.

[HenrikOlsen]

However, local processing will likely remove the suggested need for such tremendous bandwidth. Look to current MMORPGs and FPS games... they do not transmit the picture, nor does the picture need to be transmitted; only the fundamental features need to be sent then the local processor and storage intelligently assemble it into a fluid reality. Finite geometry and distinguishing features take an awful lot less bandwidth to send.

They do however require multiple DVD's worth of data installed on the local machine already . . ..

There are actually current applications where the computing and data need is such that transmitting the already rendered image/videofeed makes more sense than rendering it locally because rendering requires calculations on data picked from a dataset which is too large to have on a local system and the relevant subset of data would require higher bandwidth that a video feed of the final result does.

[cjl]

Here is a quote from the Wiki page on Bandwidth.

"Hartley then combined the above quantification with Nyquist's observation that the number of independent pulses that could be put through a channel of bandwidth B hertz was 2B pulses per second, to arrive at his quantitative measure for achievable line rate.."

For a 1 THz signal, the baud rate would be 2THz. I don't know how to explain the 1GHz rate on a 160 MHz bandwidth. My prediction would be that it would only handle .32 GHz, not one GHz.

Perhaps the device is using some form of compression.

Compression doesn't necessarily allow for an improvement in data rate, as much of the data sent by modern computers is largely incompressible (usually because it is already compressed). However, if you read the wiki page more carefully, the data rate can be much higher than 2 times the bandwidth - that's for the special case where you can only distinguish two levels of signal, on or off. If you can distinguish more signal levels, the available data rate increases with the base 2 log of the number of distinguishable signal levels. Thus, the higher the signal to noise ratio (and thus, the more distinguishable levels there are), the higher the data rate, even if the bandwidth is not increased any further.

To achieve a factor of 10 in ideal circumstances, this would require that the base 2 log of the number of distinguishable levels is equal to 5, or in other words, it requires 32 distinguishable signal levels. Of course, this is the best case scenario - in reality, they probably have some additional margin, but this is definitely achievable. Assuming I'm interpreting the equations farther down the page correctly as well, this would correspond to a minimum signal to noise ratio of about 30dB, which is definitely within reason for a short range wifi connection, even with fairly minimal transmit power.