Slashdot Mirror
Cognitive Radios Could Increase Wireless Spectrum
schliz writes to tell us that a new intelligent radio technology, dubbed "cognitive radio," is being developed that adjusts operation based on input from its surroundings. Consumers wont likely see practical implementations of this tech for another five years, but it could have wide reaching applications from wireless networking to public safety devices. "Adaptive, cognitive radios could enable techniques such as dynamic frequency sharing, in which radios automatically locate unused frequencies, or share channels based on a priority system. In public safety, cognitive radios also could be used to provide interoperability between various signals and automatically adjust radio performance."
But as with any system of resource sharing (especially bandwidth), some devices/users will simply ignore the rules to improve their own performance: flag every transmission as ultra-high-priority and so forth. You can't expect users, or even manufacturers, to "play fair." And I'm not convinced that regulation can force people to play fair. Unlike objective measures like antenna transmission power, things like "priority" are more open to interpretation (or misinterpretation, if you prefer). I suppose the same solution as for cell-phones could be applied: if you charge someone for every transmission, they are forced to conserve bandwidth.
I love technology... but when it comes to safety and emergency systems, it's usually best to use the lowest-tech solution possible: cheaper, easier to repair/maintain, more robust, more reliable, and better understood. I don't know if I want my emergency call negotiating interoperability with other devices to reach someone (since any such operation is error-prone). The simplest solution (e.g. full-power transmission on a reserved channel) is probably better in such a case.
Motorola have been working on this for years and I suspect so have all the other mobile handset makers. Frequency agile, power agile systems are used in all cell phones, base stations can direct and use a focused beam to reach a faint handset - this is in service stuff. Walk through hand off from a handset that uses DECT, then dials out and goes cellular seamlessley has been demonstrated, as have handsets that can ask others to dial power down etc. The only thing holding this back has been market forces - customers just haven't wanted it. Yet.
At least this one didn't make any "The FCC will be obsolete!" claims.
As someone else said, there's not much to prevent rogue radios from abusing the system. Also, wideband receivers are extremely difficult to design. If you design an extremely wideband receiver, you give up one of the following:
Low power
Good interference rejection
Low cost
All the DSP in the world isn't going to help you if your receiver frontend is overloaded.
retrorocket.o not found, launch anyway?
Although it seems like a neat idea, in general, the article seems to have the stink of "buzz" around it.
It's cognitive radio! It adapts, intelligently! If you go somewhere with a weak signal, your radio would get that signal some other way, intelligently! Wait, we mean broadcast towers would change their output and frequency whenever it gets crowded, and this allows things to be more crowded, but not in a bad way! The whole thing adapts, intelligently!
I would love to see details, instead of vague descriptions about how things might work. Also, who profits, and how?
Fnord.
Uber high-end (such as Xirrus) Wifi-A/P's already turn their power down when needed. What is needed is a standardized and ieee ratified protocol so that an A/P can order its connected clients to lower their power as well. This allows for packing in more and more "cells" in a given area, thereby increasing the overall throughput. It also has a nice side effect of much extended battery life. (at least in cell phones/pda's...might not be significant with a laptop.) This is nothing new, and the cell-phone industry figured it out a long time ago.
The same concept has been in play for entire existence of Ethernet, which has proven very reliable. The only significant difference here is that instead of time based - single channel- collisions where you back off for a random amount of time, you back off and frequency hop.
-- The morphemes of your disquisition are ascertainable, but they have eschewed an ambit of transpicuous exposition.
Cognitive radios!?? Oh no you don't! (Starts adding layers to tin foil hat.)
Wow! One big cloud of spectrum. Damn those oligopolistic senators who are preventing us from reaching that utopia!
Oh, and damn the people who make radio transmitters that can only recieve small slices of signal for a reasonable cost. Don't you want the cloud?!
Oh, and damn the millions of legacy devices which all require fixed spectrum blocks without negotiation. Didn't they realize that the future was coming? Damn those applications where a DSP attached to your radio is impractical.
Damn lazy americans who don't want to price-shop every time they make a phone call by looking for the reciever in their area with the lowest prices (or get trapped in an area where every reciever is gouging). Don't they realize that by checking cell tower rates every few seconds as you drive down the highway you can optimize your cash path?
But yes, damn those Senators! Why don't they just back off from this entire thing and let natural market forces take over... by writing the millions of lines of new regulations this structured market would require, including the conversion of legacy devices (and their recycling), spectrum buyback, establishing interoperative billing communications standards, testing for aeronautical interference, etc. Clearly by not jumping on the "unlimited spectrum" utopia bandwagon based upon early technological progress and conjecture, they simply hate freedom.
Why, why don't they realize that in areas untouched by federal regulation, like Operating Systems, the free market has created such perfect efficiencies? Why can't they enjoy the benefits of a truly competitive market like California's Energy markets?
The ______ Agenda
And then, we run into a problem - the standard phantom transmitter. You have three stations, A talking to B on frequency F. Then we have your station, C, closer to B than A, and as such, can't pick up A's transmission. But C starts saying "F isn't used, I'll use it!" and blasts something out. If you use the same power as A, it's unlikely that A will notice, but B certainly will.
The big problem is, this can happen a lot on the "unused" frequencies. Also, you have the issue that they may suddenly be in use at a random time, and you really shouldn't wipe out its transmissions (it may be a fleapower transmitter from some remote sensor - one way, so it can't even detect if there's interference).
To do this properly requires cooperation from everyone, and bi-directional control, which means battery powered transmitters now need receivers and processors to "slot in nicely". And radios that are somewhat frequency agile so they can move to another frequency if their primary one is jammed.
And we all know from 802.11 how well things cooperate. When it works, it works great. But then you get people who twist the spec about ("Super G", "Turbo G") so they can sell "faster!" or "more range!". Or early first-gen N devices that were reknown for wiping out the entire band by not cooperating with existing installations. Sure they conformed to the spec, provided you were on their network, but damned be everyone else. (Which is unusual, since 802.11 forces the MAC to cooperate with everyone, including those using the band but on another network, but as usual, it's an implementation problem, not a spec problem. The same happens for the 40MHz channels issue on present-gen draft-N - the radios don't listen on the sub channel, and just assume it's free when the main channel is free).
As a telecommunications engineer, I do not understand what is new with this.
If you want to check out something that will have a real impact on such video/radio applications, have a look at SVC.
I have some printet material here from this year's IBC where it seems NTT actually already have a software implementation of it.