In what world is 50 Volts/meter typical of any user near a cell site?
If Typical sector antennas have 20 dB gain, and I'm not sure they are this high for 120 degree sectors probably only 14 dB, at 20 watts average transmitter power one has to be within about 5 meters to see that sort of field strength. At 50 meters with inverse-square (far field) this falls to one hundredth that level.
Who spends significant time only 5 meters from the center of beam of a cell antenna?
I suspect that field strength from a leaky microwave oven far surpasses typical exposures from cell sites.
I think this report is BS on multiple counts.
A single satellite can easily have much more capacity than a 10 Mb Ethernet cable. Today's point-point IP radios can easily do 100 times that but a satellite's capacity is spread over a very large area. It doesn't solve the problem if you use beam forming. To cover the whole earth with N satellites, each satellite's available RF power must on average illuminate earth_area/N. That sets the best case power density with perfect patterns. In actuality it will be worse than this. On the ground, each user can not have an arbitrarily large/directional antenna. The mobile phone user wants the radio and antenna to fit in his/her pocket and run from batteries for a day.
Thus both signal power, S, and noise spectral density, N, are set. Per Shannon, this establishes a maximum average data rate per user. This is not a technology problem in that Shannon tells you what the limit is if you do it perfectly.
The only way to improve this link budget which is dictating the maximum average data rate is to add more antenna, more aperture, at the user's end and must always be low enough directivity, well formed/pointed/steered for satellite handoff. Directive antennas are inherently larger. And though they can be steered, nothing else substitutes for aperture - how big a 'bucket' to 'catch' RF they represent - which is necessary to increase S. With set ERP at the satellite the required aperture is independent of wavelength.
A LEO network *is* possible but the average per-user data rate is set by physics. I maintain that in today's market, which has an expectation of ten's of Mbps (or something similar) that the attendant per-user cost will not support the expectation. The Iridium network with (actually less than) 77 satellites was severely over subscribed. Each user could pay a lot to get a few kbps for a few minutes each day but all users could not simultaneously get many Mbps or even two way audio all the time. A few users in extreme situations were willing to pay the fee but the average user over the whole world would not be willing today.
It's physics and economics. But this doesn't keep people from investing, witness Iridium.
Do the math. First make an estimate of how much solar power your 300 (or whatever number of) satellites can catch. Then multiply that by the conversion to RF power and spread the resulting power evenly over the surface of the earth. You now have power density. Next, calculate the maximum antenna size/directivity a single user can use. His beamwidth can't be narrower than the inter-satellite angular spacing.
Next after derating the above result based on necessary link margin for foliage, precipitation loss (if it applies at the wavelength used) etc, apply Shannon's equation to this power budget and calculate the available per-user information rate.
Finally ask yourself who besides the fringe will be willing to pay enough for this relatively low average rate to support the whole thing.
As with the Iridium system, even without latency and particularly in the present age with the per-user bar up in the 10's of Mbps, the overall user base will not be willing to pay so much for so little.
For a few users the few kbps (not Mbps) average rate might be useful but it is necessary to have a lot of users to pay for it all. This is essentially a very over-subscribed approach and the physics, even with moderately good nearly line-of-sight radio paths, won't support any reasonable economic model.
The US 7B original cost of Iridium turned into something only a few tenths of a percent of that at the last sale, as I understand it. Yes, it is possible to make a system that can support a few users at high rates or a lot of users at low average rates but the economics require both simultaneously. It's not going to happen with a LEO satellite system in this day and age.
ERP is Effective Radiated Power. In the direction of maximum beam of my laser pointer, I get a spot on the order of 1 " in diameter 200' away. This means that most of the 5 milliwatts the laser puts out is contained in a spot of on the order of one square inch. This intensity is brought about by the columnation or directivity of the laser itself. It's a puny 5 milliwatt transmitter with a high gain antenna. In order to get the same intensity from an isotropic antenna (one that spews equally in all directions) rather than a directive one, I'd need to increase the power by the ratio of 4*PI*(200 feet*12 inches/foot) ^2. That's how many square inches are in a sphere with radius 200'. That's almost 80 dB (a hundred million times) change of directivity. BTW directivity is the same as antenna gain if the antenna is well matched and not lossy. 80 dB above 7 dBm ( 5 mW) is +87 dBm or +57 dBW That's HALF A MILLION watts!
But this is not "cooking power". Energy is conserved, it's still only a 5 milliwatt, Class III laser and this ERP number is only a measure of what transmit power would be necessary if there weren't any antenna gain. All this alarm about ERP is about not understanding what the terms mean. ERP is transmitter power + antenna gain, not real power. The actual transmitter is something like 24 watts, roughly the same as one segment antenna of a cell site. The system has high ERP because it's at millimeter where the antenna has a lot of gain.
This whole thread is alarm about nothing...
In spite of all the bad press *freespace* wireless isn't as terrible as you think. See http://www.corridorsystems.com... and in particular slide 16. The problem is that we live in a world with anything but freespace paths (truly laser light line-of-sight). The difference can be a factor of a million to 1 (60 dB)in throughput over common paths. Your cell phone could talk to another one 2000 miles away if you had free space but sometimes you can't get to a tower 2 miles away. Thus, this really is a problem Shannon's equation can apply to.
Wireless for 3g,4g,5g only works when the paths are *really* short - like a few tens of meters. See the rest of the paper.
what if ur in a city and the to of the buildings are already several hundred meters high? then if a thing is tethered to the ground it might bump into buildings.
Start at the building tops - cell sites already do. The issue is to get very close to LOS to the user base so that that data vs. energy is maximized. Take a look at COST231/Hata or similar real-world RF pathloss models. Anything other than LOS is a killer and can't be afforded. The present flooding model for cellular architecture is inherently broken. Montana|Idaho|etc never will get full coverage highspeed data, the present approach doesn't scale.
Going too far/high doesn't work either. Satellite distribution (from Iridium to geosynchronous and beyond) doesn't work either. Has to server too many users/needs to much devoted energy/user. Has to be lots of points of presence, LOS and close to end users.
I think he is on to something but the path lengths are too long. Presuming the market will stand for nothing less than mobility and at least 4G class data rates, physics requires that radio paths be shorter than he conjectures. Here's why:
Mobility means the user device must be powered from batteries and fit in your pocket. It must contain its own antenna. Thus there is a maximum local_storage/delivered_bit ratio available. It costs battery power to deliver a bit of information. Non-line-of-sight paths are entirely too wasteful - witness a cellphone "handy-talky" that can communicate 2000 miles in truly free space not making it into a cell site only 2 miles away in real world conditions. Wireless goes as inverse-square so that's a million-to-one loss, 60 dB.
Given one has to use truly LOS (as in laser light), the question of radio path length is answered by looking at the aperture of the antenna in the user's device. It can be thought of as a "bucket" that catches whatever falls on it. The size of the bucket is roughly the physical size of the device. While as frequency goes up (shorter wavelength) antenna gain goes up, so does path loss. The device can only catch as much flux as is falling on it. This is like solar panels - sunlight is about 1 kW/m^2 on earth- try as you might you won't get more energy per unit capture area.
If one does the link analysis and applies Shannon's equation, the ONLY solution that works requires paths of, perhaps, a few hundred meters. For this reason, the drones need to be quite low and there need to be more of them.
It turns out that this is possible if they are tethered and powered from the ground.
http://www.sonic.net/~n6gn/SWT... gives an example of a way to do this while also allowing the (heavy) network hardware to stay on the ground. Demo coming soon.
n6gn
I agree. Got to a WebSDR like http://websdr.ewi.utwente.nl:8901/ and automate the process. You can get a large amount of OTA signals to examine, in the correct ratios, styles and weightings. This requires you to decide whether or not the signal under test is CW or not but that's part of your algorithm anyway.
n6gn
I note that the OPERA report (Open PLC European Research Alliance, Document OP_WP1_D5_v0.9.doc) indicated a lot less than 200 Mbps of information capacity on typical European systems.A better guess seems to be 20Mbps on a good day. Thus, with your estimations, more like 50 kbps per home with conventional BPL/PLC techniques (below 80 MHz) seems likely.
All the more reason to move it to microwave-over-powerline.
n6gn
Don't confuse the user connection with the 'backhaul' which is the over-power-line part. However, also don't confuse 200 Mbps on a lab bench with a lot less than that over a single hop on real lines having excess noise, attenuation. The 200 Mbps hardware may only need 20 MHz of spectrum in the 4-80 MHz region to support that raw rate but after a few links are chained together throughput will likely be a LOT lower than that. Now aggregate 1000 homes onto that backhaul and you may scarcely have enough performance.
Fortunately, the smart-meter requirements for average data rate and latency are probably very small so it might all work fine - except for the ingress/egress radiation problems from the line which could be a show stopper.
Too bad they don't move it all up to microwave-over-power-line and avoid the interference problem at the same time they get 10X or more capacity improvement.
disclaimer: I resemble the above remark.
n6gn
Consider 77 satellites, each catching [100 watts] of solar power that you perfectly turn into useful, information carrying RF, and then perfectly overlay so that the entire surface of the earth is covered. That sets available flux at ground level, You can't use more gain and not lose coverage area (location independent access). Now add users with omni-directional antennas. User antennas must not only be small but generally omni-directional - they have to see all the sky and can't be high gain beams constantly pointed (too big, too expensive).
The associated antenna aperture determines captured power. Because of system noise temperature (antenna sees terra firma no matter what NF the equipment has, S/N ratio is determined, thus due Shannon capacity of link is set.
Guess what, it's not much to write home about if you plug in reasonable numbers. A few users on each satellite can get a little bit but all users can't use it all (or much) of the time.
And we haven't even talked about backhaul, real-world efficiences etc.
This problem is akin to the problem of getting 3G or 4G mobile networks to work everywhere. They don't and won't unless the paths are shortened greatly and the density of points-of-presence (cell sites) is greatly increased.
n6gn
http://www.sonic.net/~n6gn/ocar/ocar.html
This is essentially what one does with after market cellphone amplifiers, but the link offers more detail of the theory and what it takes to operate them properly. These amplifiers are bi-directional, both uplink and downlink are supported but in opposite directions. Use two isolated antennas and make the one pointed at the cell site (particularly) as directional as possible. I suggest a $50/$75 3' parabolic 'grid' reflector for PCS/850 MHz respectively. The ones offered for WiFi (2.4 GHz) actually work very well on PCS but not at 850 MHz and offer ~24 dBi gain. If you are really cheap, build corner reflectors http://www.sonic.net/~n6gn/corner.pdf.
We hear with our brains as much as with our ears. Simply buying hardware to compensate for the roll-off is NOT the solution. Hearing is tremendously adaptive and interactive. When you buy HAs from a reputable source you are actually buying a lot of visits for measurement/modification to allow you to adapt to the augmentation as well as possible. This is unique to each individual. This easily adds up to MANY (10-20/year) office visits over the life of the device(s).
I too used to think that simply measuring the roll-off and applying compensation was a solution. It emphatically is NOT.
Before you all attribute the cost of the hardware to greed, take a look at the service and also look around and find evidence of overly-fat audiologists. I don't find them around where I live...
n6gn
Most modern phones and PDAs run no more than 400 mW maximum (+26 dBm). However, that is not a typical level. Most systems utilize power control as part of the protocol. CDMA, for example, updates the channel power ~800 times per second. It is a goal of the system to use no more power from the handset than necessary to achieve parity among users sharing a system. Average transmit power may be sub-microwatt (http://www.sonic.net/~n6gn/EVDOforum/radiation.pdf
It's true some may spend more time with their phone at their ear than warming food but peak exposure from 'good' microwave ovens, never mind leaky ones, may dwarf that of communication's RF.
n6gn
I'll supply the power over a single conductive cable 1 km long if you'll supply the robot to climb it. We can share the prize. I'm ready to demonstrate. To see how I do it see http://www.corridor.biz/FullArticle.pdf
n6gn
Another way to view this fundamental wrong-tool-for-the-job issue is in considering the information rate you need to support for a phone call, by comparing a 2-way audio channel at ~10 kbps with a 3G data channel at whatever you call 3G but lots more than 10 kbps.
This "3G overhead" in both up and downlink directions requires a better radio path than for the same audio call, everything else equal. Less energy needs to be transmitted for the audio call. While protocols can play games with this fundamental fact (as does EVDO by forking over the *entire* base station carrier to one user at a time and time-slicing (oversubscribing)) the fundamental service, there's no free lunch. Your limited battery and antenna size limit the range at which you can communicate with a given hot-spot/cell-site and requires that there be higher site density to serve a given user base.
This means that a WiMax solution will fundamentally be more expensive than an old voice-only solution. In the end (whenever it all catches up with the user) this will be more expensive. Simultaneous higher data rates to all users (3G) takes more capacity/coverage than 2G. You can borrow coverage from capacity and vice-versa but someone has to pay and in the WiMax case payment will probably be initially in the form of reduced coverage area and more expensive plans. If it hasn't already hit the wall as in the case of 3G (notice how US coverage is only a few percent of the geography compared to 2G?) it will definitely do so with 4G. WiMax can throttle down to something around 1 Mbps but not a lot lower. There's 100:1 (20 dB) difference in energy delivery requirements between these two rates. This fundamental system cost is going to keep it from being an effective replacement for audio-only communications.
n6gn
I'll say! Sheesh, If your head weighs the same as a bowling ball, that means the ENTIRE output of a cell site has to be *coupled* to your head. This is easily 100,000 times more power than one would receive 50' away from a typical cell installation.
It would be 100 or more times the entire output of most handsets and probably at least 10,000 times the maximum that would likely be coupled into tissue.
I don't understand how this test is supposed to be relevant to cellphone use by anyone, anywhere.
n6gn
Hertz did a similar thing: http://en.wikipedia.org/wiki/Heinrich_Rudolf_Hertz #Electromagnetic_research probably at around 50 MHz rather than 10 MHz but pretty similar. He didn't have a 60W bulb to power at the time (Where's Edison when you need him?) but he got a *spark* at similar distance. He even made it work through a box.
I know, the difference is related to how the filed is/was generated but this certainly isn't new.
Dynamic rather than static allocation of bandwidth could indeed help both maximum rate and maximum distance for DSL. For those interested there's a plot I did of Shannon capacity for a number of different media types, (including fiber and DSL lines) at http://www.computingunplugged.com/issues/issue2006 08/00001828001.html
Note the real-world disclaimers mentioned in the text though...
While I think that a resolution that repairs the relationship is greatly superior, I don't know how to accomplish that. So here's a technical approach -- downconvert and reflect the noise back at him.
How-- connect a microphone to your sound card, preferrably a directional one that you can aim at the source of the noise. Find a sound-tools library with filters and code a highpass filter with a ~10 KHz corner. Sample the output of the high-passed spectrum at a 2 ms rate (500 Hz). Apply that output to the input of your 100W subwoofer. Set the subwoofer to have a 250 Hz upper corner, most have an adjustable setting.
Since there isn't normally much audio power in ambient settings, as long as he's "quiet" there will be no output. However when he fires up, the sampling by your system will downconvert the spectrum from 10 KHz to the upper limit of your hardware, folding it into the 0-250 Hz region where your subwoofer will enthusiastically play it back to him.
He should quickly discover the correspondence between his having the mosquito noise-maker engaged and that extremely loud and annoying frequency-translated version of it that shakes the side of his home. Not being foolish, he will turn it off at which point your system will go silent except for possibly a "burp" or two from a bat that flies through the microphones pattern.
As I said, this is really a second rate solution because it doesn't solve the fundamental problem --relationship-- but perhaps it will be interesting.
n6gn
I don't argue that a good implementation at only 1 Mbps can be better than a poor one at 11 or 54 Mbps. However, their cheerleading about smart antennas helping maximize performance with high path loss in the context of indoor antennas and (very) non-LOS paths is comparing antenna and modulation improvements of a few dB, with pathloss increases of at least 30-40 dB. Put into linear terms (as information capacity is normally expressed) that's comparing an improvement factor of a few, say 3 to 10, with path degradations of ten thousand. It reminds me of rearranging deck chairs on the Titanic as a method to improve chances for survival!
Even at only 1 Mbps, with the best of modulations and "smart" antennas, compared to the degradation on the majority of longer (1-10 mile) user paths, fixed wireless is not going to prove to be a general solution.
Use if if it can work for you in your location but don't count on it to be a widespread solution. Definitely don't invest your retirement on its technical merits in doing that.
Now, perhaps I better stop being so realistic and disparaging of Intel before I get into trouble.
Huh. Have to go now anyway, someone at the door... I wonder if those guys in black shirts driving that truck with a funny "I" on it are delivering something to me, didn't expect anything today..... (:>)
Post Comment
You are not logged in. You can log in now using the convenient form below, or
Well, I'm quite optimistic, only it's that physical laws aren't going to be violated any time soon by a standards committee, a big corporation, or anyone else who is mortal.
I glanced at Navini and might hazard a guess that the modem your acquaintances use is roughly 1 Mbps. Even at that relatively low speed (compared with WiMax claims) I stand by my assertion that 30 miles and non-LOS with laptop-sized antennas and laptop class power levels is simply not going to happen. In fact, if the modems your acquaintances have are like the "RipWave" on Navini's web site, I'm certain that if you investigate their network performance with antennas located in typical user laptop locations; indoors, shielded by buildings, trees and hills, from a 10 km distant central site, you'll discover that it doesn't even work at 1 Mbps. Probably not at all in most such locations. We're back to the vast difference between LOS/ideal sites with good (gain) antennas and non-LOS/typical locations. It's easily 1million:1 (60 dB) different and sometimes much more over multi-km paths.
WRT smart antenna and dependence upon multipath solving the problem.. is a little like saying that a company can lose money on each product they sell but make it up in volume (:>)
The link budget estimates I was making are at a level below that of the protocol. That is, I am applying Shannon's equation which basically says "if you do everything perfectly and use an optimum coding/decoding technique". I'm assuming that *all* energy, from every last path or source, is effectively demodulated as best theoretically possible. The antenn's full aperture was deemed available in every circumstance. That's another way of saying that the "smart antenna" was perfect, for it's size. My estimates were actually quite optimistic as I didn't derate things for the realities of real-world implementations like noise, interference, cable loss, coding/decoding imperfections and protocol efficiencies.
However, I take to heart your comment about following the numbers, I was generating them on the fly, as I typed, and I realize that they are probably not easy to follow for those who have been spending their time living an interesting life (:>)
Actually, it isn't necessary to look at the entire link budget to see the validity of my case. If one simply starts at the point I agree with Intel, that 30 miles *is* possible with WiFi/WiMax class hardware and the right antenna over LOS paths, all that is required is to come to an understanding of the amount typical paths deviate from ideal ones.
For this, it isn't necessary to take my word for it. Path loss modelling and analysis is a big deal for people doing fixed wireless (I guarantee that Navini has/will run into it just like the cellular carriers have--who spend a very large amount of money in site planning each year). Simply pick a model of your liking, there are several but below 1 GHz the Lee model is in favor and above 1 GHz, including PCS/CDMA-2000 and WiFi/WiMax frequencies the COST231/Hata model is often used. These models give pathloss from a (normally, high level) central site to a few "typical" user scenarios. Rural, Sub-urban, urban and dense-urban are commonly represented. For "last-mile" distances, the results are *many* tens of dB's worse than LOS. Each 10 dB is an order of magnitude of information capacity reduction.
So, by running one of these models such as the COST231 one at http://www.mathworks.com/matlabcentral/fileexc hang e/loadFile.do?objectId=2224&objectType=file which , BTW, runs OK in Octave if you happen to have a Linux distribution handy), you can pretty easily see how grim the reality is.
Columbus was right, the world wasn't flat. If it was, we might be able to actually get the kind of performance that Intel is suggesting, or allowing to be inferred from their claims. As it is, the world of the typical laptop user isn't even a smooth sphere, the top 50 meters or so where users live in most areas is extrem
A "smart" antenna, one which has gain and is dynamically steerable is certainly better for a mobile user than one with a static pattern. However, Even at 5 GHz, where there is a more directivity possible from a physically smaller antenna, any notebook I'd be carrying probably doesn't have an antenna larger than, at most, the display.
An isotropic antenna, a hypothetical reference antenna with no directivity and thus 0 dB gain, has an effective aperture of 1/(4*pi) square wavelengths. This means that a directive antenna, one with gain, has a gain of about 12 (just over 10 dB)per square wavelength of aperture. A laptop/notebook display is on the order of 30 cm x 30 cm. That's about 25 square wavelengths of area at 5 GHz. If the smart antenna phasing and steering were lossless, and the antenna were exactly broadside to the desired direction, such an antenna would have at most a gain of around 300 -- 25 dBi. In reality, it is likely to have considerably less.
BUT considering a user antenna of 25 dB gain, located in typical user surroundings and trying to use (mainly non-LOS) paths to communicate with WiMax, the results are TERRIBLE, compared to the marketing that WiMax is getting. Please see my previous post for some justification for this assertion. It is my contention, which I believe I can support by both path loss models as well as measurement, that such a laptop radio/antenna is not likely to be able to communicate even a few 10's of Mbps more than a few hundred feet. Yes, *feet* (OK, I'll modify that to 100 meters or two if you prefer those units (:>)
WiMax is not even remotely close to being able to deliver 70 Mbps and 30 miles non-LOS. It's *many orders of magnitude short* of this for any conceivable likely implementation.
Intel is selling snake oil. It doesn't have to work for them to succeed, they simply have to find buyers.
Slashdot Mirror
In what world is 50 Volts/meter typical of any user near a cell site? If Typical sector antennas have 20 dB gain, and I'm not sure they are this high for 120 degree sectors probably only 14 dB, at 20 watts average transmitter power one has to be within about 5 meters to see that sort of field strength. At 50 meters with inverse-square (far field) this falls to one hundredth that level. Who spends significant time only 5 meters from the center of beam of a cell antenna? I suspect that field strength from a leaky microwave oven far surpasses typical exposures from cell sites. I think this report is BS on multiple counts.
A single satellite can easily have much more capacity than a 10 Mb Ethernet cable. Today's point-point IP radios can easily do 100 times that but a satellite's capacity is spread over a very large area. It doesn't solve the problem if you use beam forming. To cover the whole earth with N satellites, each satellite's available RF power must on average illuminate earth_area/N. That sets the best case power density with perfect patterns. In actuality it will be worse than this. On the ground, each user can not have an arbitrarily large/directional antenna. The mobile phone user wants the radio and antenna to fit in his/her pocket and run from batteries for a day. Thus both signal power, S, and noise spectral density, N, are set. Per Shannon, this establishes a maximum average data rate per user. This is not a technology problem in that Shannon tells you what the limit is if you do it perfectly. The only way to improve this link budget which is dictating the maximum average data rate is to add more antenna, more aperture, at the user's end and must always be low enough directivity, well formed/pointed/steered for satellite handoff. Directive antennas are inherently larger. And though they can be steered, nothing else substitutes for aperture - how big a 'bucket' to 'catch' RF they represent - which is necessary to increase S. With set ERP at the satellite the required aperture is independent of wavelength. A LEO network *is* possible but the average per-user data rate is set by physics. I maintain that in today's market, which has an expectation of ten's of Mbps (or something similar) that the attendant per-user cost will not support the expectation. The Iridium network with (actually less than) 77 satellites was severely over subscribed. Each user could pay a lot to get a few kbps for a few minutes each day but all users could not simultaneously get many Mbps or even two way audio all the time. A few users in extreme situations were willing to pay the fee but the average user over the whole world would not be willing today. It's physics and economics. But this doesn't keep people from investing, witness Iridium.
Do the math. First make an estimate of how much solar power your 300 (or whatever number of) satellites can catch. Then multiply that by the conversion to RF power and spread the resulting power evenly over the surface of the earth. You now have power density. Next, calculate the maximum antenna size/directivity a single user can use. His beamwidth can't be narrower than the inter-satellite angular spacing. Next after derating the above result based on necessary link margin for foliage, precipitation loss (if it applies at the wavelength used) etc, apply Shannon's equation to this power budget and calculate the available per-user information rate. Finally ask yourself who besides the fringe will be willing to pay enough for this relatively low average rate to support the whole thing. As with the Iridium system, even without latency and particularly in the present age with the per-user bar up in the 10's of Mbps, the overall user base will not be willing to pay so much for so little. For a few users the few kbps (not Mbps) average rate might be useful but it is necessary to have a lot of users to pay for it all. This is essentially a very over-subscribed approach and the physics, even with moderately good nearly line-of-sight radio paths, won't support any reasonable economic model. The US 7B original cost of Iridium turned into something only a few tenths of a percent of that at the last sale, as I understand it. Yes, it is possible to make a system that can support a few users at high rates or a lot of users at low average rates but the economics require both simultaneously. It's not going to happen with a LEO satellite system in this day and age.
ERP is Effective Radiated Power. In the direction of maximum beam of my laser pointer, I get a spot on the order of 1 " in diameter 200' away. This means that most of the 5 milliwatts the laser puts out is contained in a spot of on the order of one square inch. This intensity is brought about by the columnation or directivity of the laser itself. It's a puny 5 milliwatt transmitter with a high gain antenna. In order to get the same intensity from an isotropic antenna (one that spews equally in all directions) rather than a directive one, I'd need to increase the power by the ratio of 4*PI*(200 feet*12 inches/foot) ^2. That's how many square inches are in a sphere with radius 200'. That's almost 80 dB (a hundred million times) change of directivity. BTW directivity is the same as antenna gain if the antenna is well matched and not lossy. 80 dB above 7 dBm ( 5 mW) is +87 dBm or +57 dBW That's HALF A MILLION watts! But this is not "cooking power". Energy is conserved, it's still only a 5 milliwatt, Class III laser and this ERP number is only a measure of what transmit power would be necessary if there weren't any antenna gain. All this alarm about ERP is about not understanding what the terms mean. ERP is transmitter power + antenna gain, not real power. The actual transmitter is something like 24 watts, roughly the same as one segment antenna of a cell site. The system has high ERP because it's at millimeter where the antenna has a lot of gain. This whole thread is alarm about nothing...
In spite of all the bad press *freespace* wireless isn't as terrible as you think. See http://www.corridorsystems.com... and in particular slide 16. The problem is that we live in a world with anything but freespace paths (truly laser light line-of-sight). The difference can be a factor of a million to 1 (60 dB)in throughput over common paths. Your cell phone could talk to another one 2000 miles away if you had free space but sometimes you can't get to a tower 2 miles away. Thus, this really is a problem Shannon's equation can apply to. Wireless for 3g,4g,5g only works when the paths are *really* short - like a few tens of meters. See the rest of the paper.
what if ur in a city and the to of the buildings are already several hundred meters high? then if a thing is tethered to the ground it might bump into buildings.
Start at the building tops - cell sites already do. The issue is to get very close to LOS to the user base so that that data vs. energy is maximized. Take a look at COST231/Hata or similar real-world RF pathloss models. Anything other than LOS is a killer and can't be afforded. The present flooding model for cellular architecture is inherently broken. Montana|Idaho|etc never will get full coverage highspeed data, the present approach doesn't scale. Going too far/high doesn't work either. Satellite distribution (from Iridium to geosynchronous and beyond) doesn't work either. Has to server too many users/needs to much devoted energy/user. Has to be lots of points of presence, LOS and close to end users.
I think he is on to something but the path lengths are too long. Presuming the market will stand for nothing less than mobility and at least 4G class data rates, physics requires that radio paths be shorter than he conjectures. Here's why: Mobility means the user device must be powered from batteries and fit in your pocket. It must contain its own antenna. Thus there is a maximum local_storage/delivered_bit ratio available. It costs battery power to deliver a bit of information. Non-line-of-sight paths are entirely too wasteful - witness a cellphone "handy-talky" that can communicate 2000 miles in truly free space not making it into a cell site only 2 miles away in real world conditions. Wireless goes as inverse-square so that's a million-to-one loss, 60 dB. Given one has to use truly LOS (as in laser light), the question of radio path length is answered by looking at the aperture of the antenna in the user's device. It can be thought of as a "bucket" that catches whatever falls on it. The size of the bucket is roughly the physical size of the device. While as frequency goes up (shorter wavelength) antenna gain goes up, so does path loss. The device can only catch as much flux as is falling on it. This is like solar panels - sunlight is about 1 kW/m^2 on earth- try as you might you won't get more energy per unit capture area. If one does the link analysis and applies Shannon's equation, the ONLY solution that works requires paths of, perhaps, a few hundred meters. For this reason, the drones need to be quite low and there need to be more of them. It turns out that this is possible if they are tethered and powered from the ground. http://www.sonic.net/~n6gn/SWT... gives an example of a way to do this while also allowing the (heavy) network hardware to stay on the ground. Demo coming soon. n6gn
I agree. Got to a WebSDR like http://websdr.ewi.utwente.nl:8901/ and automate the process. You can get a large amount of OTA signals to examine, in the correct ratios, styles and weightings. This requires you to decide whether or not the signal under test is CW or not but that's part of your algorithm anyway. n6gn
I note that the OPERA report (Open PLC European Research Alliance, Document OP_WP1_D5_v0.9.doc) indicated a lot less than 200 Mbps of information capacity on typical European systems.A better guess seems to be 20Mbps on a good day. Thus, with your estimations, more like 50 kbps per home with conventional BPL/PLC techniques (below 80 MHz) seems likely. All the more reason to move it to microwave-over-powerline. n6gn
Don't confuse the user connection with the 'backhaul' which is the over-power-line part. However, also don't confuse 200 Mbps on a lab bench with a lot less than that over a single hop on real lines having excess noise, attenuation. The 200 Mbps hardware may only need 20 MHz of spectrum in the 4-80 MHz region to support that raw rate but after a few links are chained together throughput will likely be a LOT lower than that. Now aggregate 1000 homes onto that backhaul and you may scarcely have enough performance. Fortunately, the smart-meter requirements for average data rate and latency are probably very small so it might all work fine - except for the ingress/egress radiation problems from the line which could be a show stopper. Too bad they don't move it all up to microwave-over-power-line and avoid the interference problem at the same time they get 10X or more capacity improvement. disclaimer: I resemble the above remark. n6gn
Consider 77 satellites, each catching [100 watts] of solar power that you perfectly turn into useful, information carrying RF, and then perfectly overlay so that the entire surface of the earth is covered. That sets available flux at ground level, You can't use more gain and not lose coverage area (location independent access). Now add users with omni-directional antennas. User antennas must not only be small but generally omni-directional - they have to see all the sky and can't be high gain beams constantly pointed (too big, too expensive). The associated antenna aperture determines captured power. Because of system noise temperature (antenna sees terra firma no matter what NF the equipment has, S/N ratio is determined, thus due Shannon capacity of link is set. Guess what, it's not much to write home about if you plug in reasonable numbers. A few users on each satellite can get a little bit but all users can't use it all (or much) of the time. And we haven't even talked about backhaul, real-world efficiences etc. This problem is akin to the problem of getting 3G or 4G mobile networks to work everywhere. They don't and won't unless the paths are shortened greatly and the density of points-of-presence (cell sites) is greatly increased. n6gn
http://www.sonic.net/~n6gn/ocar/ocar.html This is essentially what one does with after market cellphone amplifiers, but the link offers more detail of the theory and what it takes to operate them properly. These amplifiers are bi-directional, both uplink and downlink are supported but in opposite directions. Use two isolated antennas and make the one pointed at the cell site (particularly) as directional as possible. I suggest a $50/$75 3' parabolic 'grid' reflector for PCS/850 MHz respectively. The ones offered for WiFi (2.4 GHz) actually work very well on PCS but not at 850 MHz and offer ~24 dBi gain. If you are really cheap, build corner reflectors http://www.sonic.net/~n6gn/corner.pdf.
We hear with our brains as much as with our ears. Simply buying hardware to compensate for the roll-off is NOT the solution. Hearing is tremendously adaptive and interactive. When you buy HAs from a reputable source you are actually buying a lot of visits for measurement/modification to allow you to adapt to the augmentation as well as possible. This is unique to each individual. This easily adds up to MANY (10-20/year) office visits over the life of the device(s). I too used to think that simply measuring the roll-off and applying compensation was a solution. It emphatically is NOT. Before you all attribute the cost of the hardware to greed, take a look at the service and also look around and find evidence of overly-fat audiologists. I don't find them around where I live... n6gn
Warm up a cup of coffee in your microwave oven and drink it while you look over these measurements I made at my house: http://www.sonic.net/~n6gn/EVDOforums/radiation.pdf n6gn
Most modern phones and PDAs run no more than 400 mW maximum (+26 dBm). However, that is not a typical level. Most systems utilize power control as part of the protocol. CDMA, for example, updates the channel power ~800 times per second. It is a goal of the system to use no more power from the handset than necessary to achieve parity among users sharing a system. Average transmit power may be sub-microwatt (http://www.sonic.net/~n6gn/EVDOforum/radiation.pdf It's true some may spend more time with their phone at their ear than warming food but peak exposure from 'good' microwave ovens, never mind leaky ones, may dwarf that of communication's RF. n6gn
I'll supply the power over a single conductive cable 1 km long if you'll supply the robot to climb it. We can share the prize. I'm ready to demonstrate. To see how I do it see http://www.corridor.biz/FullArticle.pdf n6gn
Another way to view this fundamental wrong-tool-for-the-job issue is in considering the information rate you need to support for a phone call, by comparing a 2-way audio channel at ~10 kbps with a 3G data channel at whatever you call 3G but lots more than 10 kbps. This "3G overhead" in both up and downlink directions requires a better radio path than for the same audio call, everything else equal. Less energy needs to be transmitted for the audio call. While protocols can play games with this fundamental fact (as does EVDO by forking over the *entire* base station carrier to one user at a time and time-slicing (oversubscribing)) the fundamental service, there's no free lunch. Your limited battery and antenna size limit the range at which you can communicate with a given hot-spot/cell-site and requires that there be higher site density to serve a given user base. This means that a WiMax solution will fundamentally be more expensive than an old voice-only solution. In the end (whenever it all catches up with the user) this will be more expensive. Simultaneous higher data rates to all users (3G) takes more capacity/coverage than 2G. You can borrow coverage from capacity and vice-versa but someone has to pay and in the WiMax case payment will probably be initially in the form of reduced coverage area and more expensive plans. If it hasn't already hit the wall as in the case of 3G (notice how US coverage is only a few percent of the geography compared to 2G?) it will definitely do so with 4G. WiMax can throttle down to something around 1 Mbps but not a lot lower. There's 100:1 (20 dB) difference in energy delivery requirements between these two rates. This fundamental system cost is going to keep it from being an effective replacement for audio-only communications. n6gn
I'll say! Sheesh, If your head weighs the same as a bowling ball, that means the ENTIRE output of a cell site has to be *coupled* to your head. This is easily 100,000 times more power than one would receive 50' away from a typical cell installation. It would be 100 or more times the entire output of most handsets and probably at least 10,000 times the maximum that would likely be coupled into tissue. I don't understand how this test is supposed to be relevant to cellphone use by anyone, anywhere. n6gn
Hertz did a similar thing:z #Electromagnetic_research
http://en.wikipedia.org/wiki/Heinrich_Rudolf_Hert
probably at around 50 MHz rather than 10 MHz but pretty similar. He didn't have a 60W bulb to power at the time (Where's Edison when you need him?) but he got a *spark* at similar distance. He even made it work through a box.
I know, the difference is related to how the filed is/was generated but this certainly isn't new.
n6gn
I've been watching YouTube on a Treo 700P for weeks using Kinoma with Sprint and there's no content filtering or billing. Why is this suddenly news?
Dynamic rather than static allocation of bandwidth could indeed help both maximum rate and maximum distance for DSL. For those interested there's a plot I did of Shannon capacity for a number of different media types, (including fiber and DSL lines) at6 08/00001828001.html
http://www.computingunplugged.com/issues/issue200
Note the real-world disclaimers mentioned in the text though...
n6gn
While I think that a resolution that repairs the relationship is greatly superior, I don't know how to accomplish that. So here's a technical approach -- downconvert and reflect the noise back at him.
How--
connect a microphone to your sound card, preferrably a directional one that you can aim at the source of the noise. Find a sound-tools library with filters and code a highpass filter with a ~10 KHz corner. Sample the output of the high-passed spectrum at a 2 ms rate (500 Hz). Apply that output to the input of your 100W subwoofer. Set the subwoofer to have a 250 Hz upper corner, most have an adjustable setting.
Since there isn't normally much audio power in ambient settings, as long as he's "quiet" there will be no output. However when he fires up, the sampling by your system will downconvert the spectrum from 10 KHz to the upper limit of your hardware, folding it into the 0-250 Hz region where your subwoofer will enthusiastically play it back to him.
He should quickly discover the correspondence between his having the mosquito noise-maker engaged and that extremely loud and annoying frequency-translated version of it that shakes the side of his home. Not being foolish, he will turn it off at which point your system will go silent except for possibly a "burp" or two from a bat that flies through the microphones pattern.
As I said, this is really a second rate solution because it doesn't solve the fundamental problem --relationship-- but perhaps it will be interesting.
n6gn
I don't argue that a good implementation at only 1 Mbps can be better than a poor one at 11 or 54 Mbps. However, their cheerleading about smart antennas helping maximize performance with high path loss in the context of indoor antennas and (very) non-LOS paths is comparing antenna and modulation improvements of a few dB, with pathloss increases of at least 30-40 dB. Put into linear terms (as information capacity is normally expressed) that's comparing an improvement factor of a few, say 3 to 10, with path degradations of ten thousand. It reminds me of rearranging deck chairs on the Titanic as a method to improve chances for survival!
Even at only 1 Mbps, with the best of modulations and "smart" antennas, compared to the degradation on the majority of longer (1-10 mile) user paths, fixed wireless is not going to prove to be a general solution.
Use if if it can work for you in your location but don't count on it to be a widespread solution. Definitely don't invest your retirement on its technical merits in doing that.
Now, perhaps I better stop being so realistic and disparaging of Intel before I get into trouble.
Huh. Have to go now anyway, someone at the door... I wonder if those guys in black shirts driving that truck with a funny "I" on it are delivering something to me, didn't expect anything today.....
(:>)
Post Comment
You are not logged in. You can log in now using the convenient form below, or
Well, I'm quite optimistic, only it's that physical laws aren't going to be violated any time soon by a standards committee, a big corporation, or anyone else who is mortal.
I glanced at Navini and might hazard a guess that the modem your acquaintances use is roughly 1 Mbps. Even at that relatively low speed (compared with WiMax claims) I stand by my assertion that 30 miles and non-LOS with laptop-sized antennas and laptop class power levels is simply not going to happen. In fact, if the modems your acquaintances have are like the "RipWave" on Navini's web site, I'm certain that if you investigate their network performance with antennas located in typical user laptop locations; indoors, shielded by buildings, trees and hills, from a 10 km distant central site, you'll discover that it doesn't even work at 1 Mbps. Probably not at all in most such locations. We're back to the vast difference between LOS/ideal sites with good (gain) antennas and non-LOS/typical locations. It's easily 1million:1 (60 dB) different and sometimes much more over multi-km paths.
WRT smart antenna and dependence upon multipath solving the problem.. is a little like saying that a company can lose money on each product they sell but make it up in volume (:>)
The link budget estimates I was making are at a level below that of the protocol. That is, I am applying Shannon's equation which basically says "if you do everything perfectly and use an optimum coding/decoding technique". I'm assuming that *all* energy, from every last path or source, is effectively demodulated as best theoretically possible. The antenn's full aperture was deemed available in every circumstance. That's another way of saying that the "smart antenna" was perfect, for it's size. My estimates were actually quite optimistic as I didn't derate things for the realities of real-world implementations like noise, interference, cable loss, coding/decoding imperfections and protocol efficiencies.
However, I take to heart your comment about following the numbers, I was generating them on the fly, as I typed, and I realize that they are probably not easy to follow for those who have been spending their time living an interesting life (:>)
Actually, it isn't necessary to look at the entire link budget to see the validity of my case. If one simply starts at the point I agree with Intel, that 30 miles *is* possible with WiFi/WiMax class hardware and the right antenna over LOS paths, all that is required is to come to an understanding of the amount typical paths deviate from ideal ones.
For this, it isn't necessary to take my word for it. Path loss modelling and analysis is a big deal for people doing fixed wireless (I guarantee that Navini has/will run into it just like the cellular carriers have--who spend a very large amount of money in site planning each year). Simply pick a model of your liking, there are several but below 1 GHz the Lee model is in favor and above 1 GHz, including PCS/CDMA-2000 and WiFi/WiMax frequencies the COST231/Hata model is often used. These models give pathloss from a (normally, high level) central site to a few "typical" user scenarios. Rural, Sub-urban, urban and dense-urban are commonly represented. For "last-mile" distances, the results are *many* tens of dB's worse than LOS. Each 10 dB is an order of magnitude of information capacity reduction.
So, by running one of these models such as the COST231 one at
http://www.mathworks.com/matlabcentral/fileexc hang e/loadFile.do?objectId=2224&objectType=file
which , BTW, runs OK in Octave if you happen to have a Linux distribution handy), you can pretty easily see how grim the reality is.
Columbus was right, the world wasn't flat. If it was, we might be able to actually get the kind of performance that Intel is suggesting, or allowing to be inferred from their claims. As it is, the world of the typical laptop user isn't even a smooth sphere, the top 50 meters or so where users live in most areas is extrem
A "smart" antenna, one which has gain and is dynamically steerable is certainly better for a mobile user than one with a static pattern. However, Even at 5 GHz, where there is a more directivity possible from a physically smaller antenna, any notebook I'd be carrying probably doesn't have an antenna larger than, at most, the display.
An isotropic antenna, a hypothetical reference antenna with no directivity and thus 0 dB gain, has an effective aperture of 1/(4*pi) square wavelengths. This means that a directive antenna, one with gain, has a gain of about 12 (just over 10 dB)per square wavelength of aperture.
A laptop/notebook display is on the order of 30 cm x 30 cm. That's about 25 square wavelengths of area at 5 GHz. If the smart antenna phasing and steering were lossless, and the antenna were exactly broadside to the desired direction, such an antenna would have at most a gain of around 300 -- 25 dBi. In reality, it is likely to have considerably less.
BUT considering a user antenna of 25 dB gain, located in typical user surroundings and trying to use (mainly non-LOS) paths to communicate with WiMax, the results are TERRIBLE, compared to the marketing that WiMax is getting. Please see my previous post for some justification for this assertion. It is my contention, which I believe I can support by both path loss models as well as measurement, that such a laptop radio/antenna is not likely to be able to communicate even a few 10's of Mbps more than a few hundred feet. Yes, *feet* (OK, I'll modify that to 100 meters or two if you prefer those units (:>)
WiMax is not even remotely close to being able to deliver 70 Mbps and 30 miles non-LOS. It's *many orders of magnitude short* of this for any conceivable likely implementation.
Intel is selling snake oil. It doesn't have to work for them to succeed, they simply have to find buyers.