Slashdot Mirror
UWB Wireless Access Could Be Here Soon
fluppy88 writes: "802.11b doesn't have anything on UWB. With a potential of 1000M bits/sec it blows the pants out of 802.11b and doesn't eat up the tightly controlled spectrum. This article on CNN gives an interesting introduction to UWB, another candidate in the future of wireless." It was mentioned here a while ago, but much more mired in controversy about whose idea it was. Now there are several companies which seem anxious to get products based on UWB to market -- if it's approved.
I wonder what that means? The FCC is going to kill this quick if it messes with GPS, which is IMO, more important.
All of the subjectivity in (how one defines unacceptable) will probably cause everyone to take a few more years before they start doing anything with this.
-- Dan
For one thing, because UWB pulses don't actually use a traditional radio signal, called a carrier, UWB transmissions don't take up any of the radio spectrum. Spectrum is limited, and demand for it is growing fast. That's one reason for the FCC interest: UWB would allow a whole new class, and volume, of voice and data communications that, in effect, wouldn't take up any more "space" in the crowded radio spectrum.
But there is concern that UWB transmissions, especially for UWB devices that will operate below about 2 GHz, will interfere with other broadcasts. These include the Global Positioning System (GPS), public safety nets, air traffic, marine navigation and communications, AM and FM radio, and television broadcasts, to name just a few.
Where do they get these guys? First he says that it doesn't use any spectrum...then he says that anything below 2 GHz will interfere with existing Nav and Comm systems. Gotta be one or the other. Can't be both.
(BTW pulse transmissions do take up spectrum, even if they don't have a carrier...)
You're using her as bait, Master!
Another technical and practical article at Intel - Ultra-Wideband Technology for Short- or Medium-Range Wireless Communications
...and IF it is approved outside the US as well!
This is not unimportant. Prices drop and rapid adoption increases when a standard is worldwide (like 802.11b on 2.4 GHz).
The 5 GHz equivalent of 802.11b (.a) will be approved at the world radio freqeuency conference in 2003 (light speed for governments) - and I was already told by the British govt Radio Agency
that the UK frequency will differ slightly from the US frequency. And that the 5 GHz standard wil be approved for commercial use (unlike the current 2.4 GHz standard).
That's just for one country, the UK. Imagine when all others (Japan, Europe, etc) also get in on the act. Result: nothing moves.
So, nice as all these new 'standards' are, I am afraid they will slow down wireless adoption.
---
BDOS ERR ON A:>
"These systems can use 50 to 70 milliwatts of power," says Adrian Jennings, technologist with Time Domain in Huntsville, Ala., one of the pioneer vendors in UWB. "That is one ten-thousandth the power of a cell phone."
50 to 70 milliwatts is one ten-thousandth the power of a cell phone? That would place an average cell phone at somewhere around 600W.
Strange, I don't remember seeing huge heatsinks and 12" fans on any cell phones lately.
Tarsnap: Online backups for the truly paranoid
This UWB term I've never heard of, but I worked at a company that was developing 802.11a goods. This sounds like the same, as the article touts ~40-55mb/ps (I don't know what this 1000MB/ps shit is).
:)
I wasn't one of the engineers working on it (I was actually a high school co-op who worked on higher-level code in the same dept), so some of my facts may be off. 802.11a (or at least the variant we were working on) used a modulation scheme called OFDM. OFDM was "invented" in the 1960s if I remember, but the technology is finally catching up with the math to allow for mass production and the data precision required in the algorithm.
OFDM would fit with the articles blurb about it being in the "noise" area. Basically, a baseband signal is multiplexed into multiple low-power subcarriers, which are aligned in such a way that the intersymbol/intercarrier interference (ISI/ICI) is minimized. Basically, this means orthangonally (at 90 deg angles), so that the peak distabution of one carrier occurs at the zero points of the carriers on either side of it. So it's a particulary advanced form of FDM. All that low power shit comes from this fact, and that the nature of noise is amplitude-related, not frequency related. Plus, data interleaving and error detection coding (described below) goes on during baseband processing I think. I forget the symbol length and all that in detail crap, but there is QAM coding and FFTs/IFFTs going on in the process. (I remember 64-QAM being a popular initial choice.) Error correction/detection might be left open in the specs (i.e. it could be this or that), but the one I was familiar using was reed-soloman (a convolutional encoding method used with CDroms) and/or turbocoding (a very advanced convolutional encoding method which gets pretty close to the limit imposed by the Shannon theorem).
OFDM has been defined as packing the data as close as physics will allow, and it whoops 802.11b in both range and bandwidth. I think it will be both in the 2GHz and the 5.4GHz bands.
Sounds exciting. The race is on.
I have seen a good intro paper on OFDM before, but I lost the URL, here is a more indepth one on it: http://www.eng.jcu.edu.au/eric/thesis/Thesis.htm
Sorry about the spelling, I'm not using a spell checker.
UWB is a clever form of spread spectrum technology. Spread spectrum (SS) is defined as any radio communications system in which the occupied bandwidth is much wider than the baseband (information rate) signal. The most common forms of SS include FM broadcast radio, where 200kHz is used to transmit about 50kHz of signal, and CDMA - cellular phone spread spectrum invented by Qualcomm. Spread spectrum was actually invented by the actress Heddy Lamar for use during World War II and was used for secure communications between Roosevelt and Churchill.
Spread spectrum has a parameter called "spreading gain" which is the ratio (expressed in DB) of the occupied bandwidth to the baseband signal. UWB is a form of spread spectrum with an extremely high spreading gain - it occupies a whole lot of spectrum (contrary to some claims) to transmit a relatively small amount of information. However, because the signal is spread over a very wide frequency range, very little signal is required on any given frequency (or technically, in any given narrowband channel). Thus the signal appears to ordinary receivers as an increase in background noise, and under most circumstances will not do so in a noticeable way.
Traditional spread spectrum uses one of two modulation techniques to mix the information signal with a spreading signal: direct sequence (DSS) and frequency hopping (FH).
Direct sequence uses a bandwidth constrained random noise generator (typically a pseudo-random digital bit stream) and multiplies the baseband signal by this. It is also band limited, either/or by filters or the spectral characteristics of the pseudo-random noise. DSS is used in CDMA cellular phones.
Frequency hopping involves moving the carrier frequency frequenly, typically in a pseudo-random manner. In fact, usually the frequency changes a number of times for each bit transmitted.
Both techniques allow reception of the transmitted information by synchronous detection - the spreading signal is duplicated in the receiver, and used to recover the baseband signal. In the case of DSS, you generate a precisely timed replica of the transmitter's pseudo-random sequence, and multiply it by the input from the antenna (or in the intermediate frequency stages - dependinng on receiver design). Low pass filtering (integration) of the output yields the original signal.
Spread spectrum systems have some or all of the following characteristics:
To get back to UWB, it uses very narrow pulses as its spreading signal. The Fourier spectrum of a very narrow pulse shows a very flat distribution of energy over a very wide bandwidth. In this sense, UWB is spread spectrum. Likewise, it recovers the signal in a similar manner as other spread spectrum signals - it uses a regenerated narrow band pulse to synchronously sample (a form of multiplication) the radio spectrum, thus recovering the original pulse (minus pulse spreading caused by reflections and frequency dispersion).
AFAIK one could duplicate the behavior of a UWB system by using an extremely wide band direct sequence system. It would provide the precise ranging, see-through wall radar characteristics. It would have the low detectability. It would have the low interference to narrower-band signals. However, the UWB system appears to be much easier and inexpensive to build.
The only good weather is bad weather.
Yes, it can be both
No it can't.
There is only so much spectrum. The faster you change the signal, the broader the chunk of spectrum you use.
You can use it for a shorter time, and end up with the same time-bandwidth product.
You can control signal intensity more finely to encode more bits, until the ambient noise (or weaker interfering signals) would confuse the decoder.
You can direct your signal so that most of the energy goes toward the receiver rather than spreading out uniformly (though this gets harder as the bandwidth gets wider).
You can restrict its polarization to one of a complimentary pair, leaving the complimentary polarization's half of the spectrum free (or using it for a second or a return signal).
... A radar (with its very directional antenna) will show only a wedge of light in the direction of your transmitter. (Which is how the authorities will find you to shut you down - if they get there before the neighbors with the torches and pitchforks.)
But that's IT.
If two transmitters are hitting a receiver with energy in the same chunk of spectrum they interfere. Spread the actual bits around so they're transmitted redundantly in different parts of the spectrum (not just hop the carrier to put a burst of bits in one chunk and move on) and you might be able to pull them out of some interference. But then you've used up several times as much spectrum in the first place - as have the interfering signals from other users of the same scheme.
UWB works by sending single-cycle pulses. The information is carried by when the pulse is transmitted with respect to a reference.
Since there is no carrier, it doesn't affect a specific part of the spectrum. However, since there is no carrier, it affects all parts of the spectrum by adding to the noise floor. That is what the big problem with this technology is and why the FCC is looking so closely at it. The UWB Consortium [uwb.org] has more information.
This scheme doesn't "not use spectrum". It uses the ENTIRE spectrum - up to the limit of the transmitting equipment. The shorter the pulse (so you can space them more closely and send more bits) the higher the limit. When it talks it steps on EVERYTHING - one pulse, one "POP" in a radio, one fleck of snow on a TV screen, one bright spot on a radar. The has to be above the noise floor itself to be heard - and the "noise floor" includes all those other signals it's interfering with.
If your data rate is low you can keep the signal weak - at the receiver. The pulses spread out their energy over an enromous chunk of bandwidth, so your reciever can measure the energy over the whole specrtum and recover the desired data from the other signals-assuming there is only ONE transmitter using this scheme, of course.
But electromagnetic signals fall off with inverse-square. Near the transmitter you're not just "raising the noise floor". You're generating a continuous lightning bolt.
Want to send data a hundred miles? Imagine you were doing this with VISIBLE light. You're modulating an arc light at your transmitter, bright enough that the output of a solar cell a hundred miles away has more signal from your arc light than from all the other light sources (including any other arc lights) combined. Now imagine somebody on your block trying to read morse-code flashes from a distant colored lightbulb, using his own solar cell and a color filter.
Run a protocol so the radio versions of your "arc lights" take turns and you can run a network. (Think of running "Ethernet" in the real aether.) But that's ALL you'll be able to run. Goodbye TV, goodbye AM and FM radio, goodbye aircraft band, police band, CB,
Meanwhile, just as the time-limited signal of time-domain modulation schemes selectively "punch a hole" in the time distribution of the signals of the frequency-domain modulation services, the frequency-domain modulation signals selectively punch holes in time-domain modulation signals. The "hole" is in the form of pattern sensitivity - selective interference with those bit patterns that correspond to the energy distribution of the frequency-domain signal. Your pulse strength has to be great enough to "shout down" this interfering signal, or those bit patterns just don't go through uncorrupted.
Time-domain and frequency-domain signals don't play well together. And the time-domain kids got to the playground first. Do you want ANOTHER war with broadcast media? Remember, if they lose it's their death, so they'll fight REAL HARD.
You can avoid the war by shaping your pulses so the energy stays in a limited band (at the cost of limiting your data rate correspondingly). But within that band you can only use time-domain schemes. You've "divided the playground". And the smaller the hunk of playground you got, the lower your data rate. Shaping your pulses stretched them out - and you have to move them farther apart to tell them apart at the receiver. How much playground do you think you can get for your gang's exclusive use?
Personally I don't see a problem with raising the noise floor for this technology because, as I understand it, it raises the floor uniformly and, if I understand this correctly, the actual number of devices transmitting doesn't play into this.
Each one of those "arc lights" raises the noise floor - by a bunch. More noise means you can't measure the signal from the desired "arc light" as accurately - which means you get less bits-per-second from it.
The total number of bits-per-second available to ALL transmitters at any given receiving antenna is a constant described by Nyquist: 2 * bandwidth in cycles-per-second * base-2 log of the signal-to-noise ratio.
But the distribution in space of the "raised noise floor" is a face-down morning-glory flower. Like those surfaces where they roll a ball bearing to demonstrate orbits, but upside-down.
Imagine a rubber sheet: You grabbed it at your transmitter and stretched it WAY up - until the sheet at the distant receiver was raised enough that the receiver could detect you yanking. And your neighbor's TV antennas are up on the peak with you. (And so is your network antenna...)
The only thing I don't quite grok is how they can get two devices to have such rock-solid stable time references (we're talking sub-picosecond jitter) without secondary clock transmitters and keep them that way. If anyone out there can help shed some light on it I'd love to hear from you.
You sacrifice a part of your bandwidth to send a pre-defined, typically repetitive, synchronization signal, to keep the receiver synchronized with the transmitter. Think of the start/stop bits on a serial line, the framing bits in T1, T3, or SONET, or the vertical and horizontal sync bars in a TV signal.
The more bandwidth you sacrifice, the faster your receiver's clock syncs up when reception starts. If you're transmitting bursts you can put most of the sync at the start of the burst to get things locked quickly (and identify the signal and its start), then use just enough to keep the receiver locked for the rest of the burst.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way