Software-Defined Radio Could Unify Wireless World

mjdroner writes "Technicians in Ireland are testing a device capable of skipping between incompatible wireless standards by tweaking its underlying code. The article states: 'The device can impersonate a multitude of different wireless devices since it uses reconfigurable software to carry out the tasks normally performed by static hardware. The technology promises to let future gadgets jump between frequencies and standards that currently conflict. A cellphone could, for example, automatically detect and jump to a much faster Wi-Fi network when in a local hotspot.'"

Comments

Software radios a step towards real deregulation?

[dada21]

I've been researching similar technologies over the past few years because I believe we can see an amazing communications "utopia" by deregulating (or at least minimizing regulations) all the frequencies we're blocking for specific uses.

Software radios are not new technology, but the implementation has been fairly worthless as frequencies are set up for specific purposes. At any given moment in any given area, there is a ton of bandwidth going unused. Frequency hopping is already pretty well documented in how to maximize its use, and power allocation specifications have been out since pre-cell phone days. Combine that with a much wider bandwidth and we can see higher data rates, lower battery usage and maximum bandwidth allocation everywhere you go.

I know the FCC will never give up the bandwidth to the open market -- it is too lucrative for the few who are in cahoots with the licensing body. But I see so much happening just in the WiFi "unreglated" spectrum that I would really love to live in a world where all that analog TV, digital TV, analog radio, digital radio, CB, HAM and every other heavily regulated piece of spectrum could be allocated to being used for just information transmission. Software radios would set themselves to the best frequency possible to maximize transmission distance (as needed) and minimize power consumption (as needed).

What we have now is more kludge than efficiency. Can you imagine how incredible the Internet would be if we had nearly infinite spectrum to use (compared to the limited spectrum we have now)?

Sure, some people will say "What prevents Megacorp YYY from blasting 100,000 watts over every frequency?" That's pretty simple -- energy costs make it prohibitive to transmit anything but profitable data. The FCC has existed long past its useful life, maybe it is time to open up little bits of unregulated spectrum piece-by-piece and let's see what happens. These software radios are a huge step in the right direction.

Software Defined Radios

[squidguy]

What's the big deal here? The US military and public safety sectors have been using radios with software defined waveform capability for over a decade. Expensive, but Moore's Law will drive the cost down to make devices using this technology commerically viable.

antennas?

[adrianmonk]

I'm not much of a hardware person, so maybe I'm missing something, but aren't different types of antennas needed for different applications? Isn't the best size of antenna a function of its frequency? I understand how you can use software to replace some of the active circuitry, but how are you going to change the size and shape of an antenna via software?

Re:antennas?

[SuperQ]

Antennas only affect the shape of the RF output, and the frequency range at which you can efficiently radiate.

Take some examples:
Omni-directional wifi antennas on most APs: a single stick with a fraction of the wavelength of 2.4ghz. Very simple, can do anything from SSB, AM, FM, or OFDM modulation.

Most of what software defined radios is talking about modulation changes, not frequncy changes.

The only difference between 802.11b and 802.11g is the modulation (CCK vs OFDM)

Re:antennas?

[satguy]

I understand how you can use software to replace some of the active circuitry, but how are you going to change the size and shape of an antenna via software?

You don't necessarily need to change the size, you need to change the resonant frequency and impedance. This is currently done with 'automatic' antenna tuning circuits using varicaps and other components/switching circuitry that varies the resonant frequency by varying bias voltages.

GNU Radio

[mukund]

Perhaps GNU Radio is of a worthy mention here.

The example you gave could kill free wifi.

[artifex2004]

Can you imagine what will happen when people find the free hotspots they're providing are constantly saturated by cell phone connections? At least until they firewall whatever protocol a phone would be using to wrap the equivalent of VoIP.

A plug for GNU Radio

[InterruptDescriptorT]

Since we're talking about Software Defined Radio, I urge everyone with an interest in the subject to look at the GNU Radio project. They have designed a front end board using generic cable TV tuners feeding an FPGA to perform some initial processing, such as decimation and filtering. The data is then transferred over USB to the host, whose software performs the demodulation and decoding. It's a fascinating project and a great stepping stone into the field.

To really get started on SDR, check out the Ten-Tec RX320D shortwave receiver. It outputs a 12 kHz-wide IF signal from the front end to an audio jack, which can then be fed to a PC soundcard. There are a number of packages that can take this data and demodulate it, including DREAM, an open source DRM (Digital Radio Mondiale) decoder which allows you to listen to the new digital shortwave transmission standard that many of the world's broadcasters are beginning to experiment with.

Re:A plug for GNU Radio

[evilviper]

To really get started on SDR, check out the Ten-Tec RX320D shortwave receiver. It outputs a 12 kHz-wide IF signal from the front end to an audio jack, which can then be fed to a PC soundcard.

Why spend that much ($350+), when you can order a dirt-cheap shortwave radio for maybe $40 and just use a simple 455 kHz to 12 kHz adaptor?

Re:A plug for GNU Radio

[leighklotz]

Why spend that much ($350+), when you can order a dirt-cheap shortwave radio for maybe $40 and just use a simple 455 kHz to 12 kHz adaptor?

SDR is a broad topic. Wide-band digital modes such as the 12KHz wide DRM or even narrow ones such as HamDream are a simple example.

SDR involves a variety of techniques, but the basic idea is using an A/D at an early stage, and performing operations traditionally done with RF components with DSP software instead.

In its extreme, an SDR has a broadband RF amplifier and a DSP.

Some systems use a tuned RF filter before the RF amplifier to improve dynamic range and reduce overload, and others put the DSP after the first analog mixer. Ham equipment that uses IF DSP does this, such as many of the ICOM radios.

Then there are devices that then mix down to somewhere around the audio range, at least to the 0-96KHz or 0-48KHz range handled by many popular PC sound cards. The RF signal is detected by a an I-Q detector, which produces two signals In Phase and Quadrature (90 degrees out of phase). You might notice that this is a decomposition of a periodic wave into real and complex parts, given v=cos(omega)+j sin(omega). Thus, DSP techniques such as FFT can be applied in the complex domain. If you're seriously interested in this math, look up the Hilbert transform. It lets you modulate or demodulate directly in the DSP, and as a result the transmit and receive software and hardware are very similar. (And wouldn't the Professor on Gilligan's Island like to know that you can make a receiver into a transmitter without using coconuts!)

Anyway, once you get the I-Q signals into the two channels of the sound card, you get a view of the RF spectrum all at once, up to the bandwidth of your sound card sampling. So, if you have a 48KHz sound card you get 48KHz of band scanned simultaneously, and can pick and choose what frequency you want to demodulate, and how you want to demodulate it in software (AM, Single-Side Band, FM, various digital modes such as the aforementioned DRM=digital radio mondial). See here and here.

The SoftRock 40 and its replacement, the SRv5, surface mount kits costing in the $30 range, do this. They're an excellent introduction to SDR techniques, without requiring DSP chip programming. People are doing fun things with them. It's not a transmitter yet, but it will be soon with another board and a ham license).

For software, among others, there is Gnuradio, and also SDRadio, a Windows app. And there's DTTSP, a SourceForge project that runs in Linux and also releases a DLL used by the FlexRadio people. DTTSP has a number of front ends in development, in Java and other languages.

A step up is the FlexRadio SDR-1000, alluded to above. It's a 100W transceiver that does the same thing that the SoftRock does, but does a better job, and also use a VFO that allows it to pick what frequency range it operates on, rather than being limited to a particular crystal-controlled band as the stock SoftRock does. It also costs quite a bit more, and they use a 96KHz sound card to get good quality.

Great

[TubeSteak]

Once upon a time, it used to be that only the military could fark up my garage door opener.

Now everybody will be able to.
Thanks Ireland.

Re:Great

[jc42]

[I]t used to be that only the military could fark up my garage door opener. Now everybody will be able to.

Heh.

However, to put it in perspective, we should note that this is directly in line with the original design, back when the Internet was called ARPAnet.

The funding came entirely from the US Dept of Defense, and if you dig up the early ARPAnet docs, you'll find lots of diagrams of military scenarios, with everything communicating via wireless links. This makes sense, of course, because you really can't tie together tanks, jet fighters, aircraft carriers, etc. with wires.

If you read the docs, you'll find that there was a strong emphasis on automatic reconfiguration, as the enemy shot down your comm equipment. Routes were to be reconfigured dynamically. The network was to use whatever comm equipment was available. It was to use whatever frequencies were usable to get the data through.

But primarily, as things got shot down, everything was supposed to constantly monitor the electronic environment, and dynamically reconfigure itself so that it kept working.

Now it's four decades later, and people are coming up with the same ideas, and pushing them as something new. Except now the "enemy" is the FCC and the corporations that want to control their part of the spectrum and block access to the competitors' equipment.

Maybe some day we'll actually get what was conceived back in the 1960's.

Re:Great

[jcr]

Once upon a time, it used to be that only the military could fark up my garage door opener.

Only the military could buy batteries, transistors, inductors, capacitors, and resistors in your country?

Sounds pretty harsh.

-jcr

Re:Great

[jc42]

We could create an army with self-routing radio communications. But it'd be expensive as all hell.

It would only be expensive to develop. That requires a bunch of smart programmers who are willing to go against the commercial grain. But it's really just a SMoP (Small Matter of Programming. Once implemented, the hardware wouldn't be materially more expensive than the ad-hoc mess that is currently in use.

The real barrier has been the same all along: Commercial suppliers have a strong incentive to try to block communication with the competitor's equipment. This was a large part of why ARPAnet was funded in the first place. And commercial obstructionism still has the upper hand.

To get a truly capable comm system that works with all vendors' equipment and is resiliant to equipment failures (e.g., from a missile strike, an ISP's QoS tactics or a government's censorship) requires some smart programmers who aren't in the pay of any of the vendors.

But once implemented, software can be replicated almost for free.

Re:Great

[jcr]

I think we can all agree on common-sense technology control, such as banning "assault capacitors" with a rating higher than 100uF.

You can have my 10 Farad capacitors when you pry them from my cold, dead hands.

Make sure they're fully discharged first, of course..

-jcr

Here's the big deal

[TubeSteak]

The underlying technology has the potential to revolutionise wireless communications but has been difficult to test outside the laboratory until now as the majority of the radio spectrum has already been allocated. Licences are normally limited to a particular radio frequency and modulation but the one issued to CTVR permits a device to hop quickly between many different standards.

The CTVR trial will also test how easily frequencies can be dynamically allocated to different devices. One idea is for companies that own a licence to automatically "sublet" access depending on demand. The licence [to trial the software radio] means we will be the first research centre in the world to practically investigate the commercial potential of dynamic spectrum-allocation,

Ever wonder why your WIFI card needs firmware?

[gaijin_]

It is because it is basically a software defined radio. You have a DPS and a set of AD/DA converters and a baseband (low freqency) to RF (high freqency) converter.

Only difference here is that they are hacking the firmware for their Atheros wifi cards a bit more than the rest of us.

Re:Software radios a step towards real deregulatio

[jcgf]

I think you need to remember that the RF spectrum is a finite resource and as such needs to be managed to prevent noise. Deregulate and you would have people lusing HF to go across town and I didn't pass the ham exams to have some bitch in kentucky's email to her boyfriend blocking DX when he lives down the street.

But you have many problems to fix first...

[AB3A]

First, how are you going to get the dynamic range you need for RF intense areas if the radio front end has to remain wide open for octaves?

Second, energy costs for radiating Kilowatts are relatively cheap.

Third, what about the near-far problems with spread spectrum?

Fourth, how do you regulate narrowband emergency frequencies in a spread spectrum world?

Fifth, if you're going to push everyone to unlicensed spread spectrum, how do we resolve interference disputes?

I could go on, but I think you can figure out where I'm coming from. The problem is that if we didn't have radio and we were starting from scratch, you might be able to make a case for this technology. But since you clearly don't know how the standards got to where they are today you have no technical basis for trashing them.

Keep dreaming until you get a clue...

Re:Software radios a step towards real deregulatio

[LostCluster]

Sure, some people will say "What prevents Megacorp YYY from blasting 100,000 watts over every frequency?" That's pretty simple -- energy costs make it prohibitive to transmit anything but profitable data. The FCC has existed long past its useful life, maybe it is time to open up little bits of unregulated spectrum piece-by-piece and let's see what happens. These software radios are a huge step in the right direction.

That's just nonsense. With no regulation at all, no frequency would be safe from a fly-by-night operation disrupting service. TiVos would fly right out of the window because they're all dependant on a show being broadcast on a designated channel at a designated time. If TV channels could be jammed by absolutely anybody, TV as we know it would cease to exist. Radio as we know it would cease to exist. DirecTV, Dish Network, Sirius, and XM would all have their investments in satellites wasted. (Only cable would be safe, because cable network signals spend most of their distance traveled on fiber anyway, and they can sheild the "last mile" of coax against anything given a thick enough wire.)

Of course, your answer to that would be that's the whole point, you want the existing license-holders evicted to free up everything for one big utopian mesh network. That's great on the chalkboard, but fails to stand up to real-world attacks. Just how are you going to avoid man-in-the-middle attacks and eavesdropping? "Encryption!" you say? That's nice, but you can't encrypt routing data. You have to put the address you're headed to outside of the enveleope, otherwise it's not going to get there. Basically, every partisipating router is a chance to be logging, and worse yet tampering with every packet. Meshes don't work as soon as somebody intent on not playing fair joins.

You basically want to obsolite nearly every RF telecom device in existance, and in that system's place subsitute chaos. It's just not gonna fly. Don't bother.

Easy to see...

[rscrawford]

It's easy to see where this will go.

1. They'll be successful, and the world will benefit from their tool.

2. Some two-bit company from Podunk, Indiana, will claim that they have a patent on the technology. A lawsuit causes the court to issue an injunction against using software radio.

3. Some other two-bit company from South Podunk, Iowa, files a suit claiming that software radio diminishes trade opportunities. The US government agrees and bans the technology. They try to get the EU to ban it as well, and a tussle ensues.

4. Large corporations take over the technology and introduce a tiered system of access.

5. Microsoft says they were planning it all along.

In the end, no one benefits from the groundbreaking technology.

Or, at least, that's how these things seem to be going these days.

Did I mention the patent infringement lawsuit?

Re:Software radios a step towards real deregulatio

[SagSaw]

Sure, some people will say "What prevents Megacorp YYY from blasting 100,000 watts over every frequency?" That's pretty simple -- energy costs make it prohibitive to transmit anything but profitable data.

No, it's not practical to blast 100,000 watts over ever frequency. I'm not worried about such a shot-gun approach. What I'm much more worried about is the "sniper" approach. Let's say that you're using the newly-deregulated spectrum to provide some service. Perhaps your trying to operate a local public interest radio station, or providing internet service, or selling wireless telephone service of some sort. Now somebody with an interest in preventing _you_ specifically from providing your service comes along. Maybe they don't like the message from your radio station, or are your competitor in the ISP/phone market. All they have to do to keep you off the air is tranmit a signal which degrates the SNR of your signal sufficiently to render it useless. If they're willing to pay a little more for the power to produce their signal then you are for the power to produce your signal, they'll win.

I'd much prefer minimal regulation (i.e. just enough to force licensees to co-operate to avoid and resolve cases of interfearance.) to no regulation.

On the list of things that make me go hmm....

[Excelsior]

The technology promises to let future gadgets jump between frequencies and standards that currently conflict.

So, in one fail swoop they've automated the radio dial and the AM/FM button? Science rocks.

Power consumption

[s!mon]

It would be nice except Slashdot has been posting something about software radio for years and still nothing. Other factors to consider:

1) Power consumption on software radios will be much less efficient than their analog counterparts
2) band limited to certain frequencies - relevant because higher powered transistors at higher frequencies are becoming available, pushing beyond the 2.5 GHz limit we have right now (compliments of Gallium Nitride and Gallium Arsenide).

Re:Software radios a step towards real deregulatio

[evilviper]

I've been researching similar technologies over the past few years because I believe we can see an amazing communications "utopia" by deregulating (or at least minimizing regulations) all the frequencies we're blocking for specific uses.

Same old anarchic anti-government trolls, huh dada21?

At any given moment in any given area, there is a ton of bandwidth going unused.

Yes, because of all the legacy equipment still in-use. You can't phase it out overnight can you? Everyone is bitching that the FCC is forcing broadcasters to shut-off their TV signal, all the while saying the FCC should be much quicker in forcing OTHER PEOPLE to replace all their radio equipment... no, that's not hypocritical at all.

Actually, it's quite ironic that the organization you want to do-away with has, in-fact, been the force gradually making companies reduce their spectrum use through newer technologies.

Combine that with a much wider bandwidth and we can see higher data rates, lower battery usage and maximum bandwidth allocation everywhere you go.

Sure, if you take the bandwidth away from somebody else, and use it for your own purposes, you'll get better data rates...

And "maximum bandwidth allocation is just a euphamism for noise all across the spectrum...

Can you imagine how incredible the Internet would be if we had nearly infinite spectrum to use (compared to the limited spectrum we have now)?

No, I can't imagine it being all that wonderful really. How does that solve any of the problems of the internet? Bandwidth would be somewhat cheaper, and you wouldn't be tied-down to a landline, but that wouldn't stop spam, trolls like yourself, DDoS attacks, worms, etc. It wouldn't really increase the content on the web, make it more useful, or make it more accessible to people in poor nations.

maybe it is time to open up little bits of unregulated spectrum piece-by-piece and let's see what happens.

While not completely unregulated, the CB band is wide-open, and getting practically no use. Before you start complaining that you don't have enough spectrum for public use, why not try utilizing what is available first?

Cap max. wattage allowed & ban delbrt interfer

[Anonymous Coward]

What prevents Megacorp YYY from blasting 100,000 watts over every frequency?

Well miscreants may do it .. or well political groups may do it to silence broadcasts etc.

It's better to simply ban intentional misuse and place a wattage cap for license free broadcast. Also open up more spectrum for WiFi and devices that are non parasitic (fine people who don't follow spread spectrum rules etc. if they are broadcasting above a certain wattage).

I would really love to live in a world where all that analog TV, digital TV, analog radio, digital radio, CB, HAM and every other heavily regulated piece of spectrum could be allocated to being used for just information transmission.

That's the way it is today. Except a corporation known as government owns it and uses / rents it out to make money. Shareholders in govt. are called "taxpayers". You are pro entity ownership right?

Cognitive radio-SDR

[slashdotmsiriv]

Cognitive radio is a concept very related to this discussion. I googled a little about SDR and cognitive radio and came across some interesting paragraphs. Short Definition of SDR: Software defined radios are making it possible to change waveform properties and applications while operating in the field via the addition or upgrade of software. For SDRs, reprogramming or upgrading a single radio or a radio network takes about as much effort as upgrading a computer's operating system or program options. US Army interest : For its part, the U.S. Navy is likely to be the largest consumer of software defined radios with the military's Joint Tactical Radio System Initiative (JTRS) radios following closely behind. For the Navy, the software-based Digital Modular Radio (DMR) is replacing a roomful of radios with a single rack of DMRs. The DMR is a four-channel, full-duplex system that is essentially four radios in one. Currently operating on submarines and surface ships around the world, the DMR (AN/USC-61) is successfully demonstrating the viability of software defined radios on active duty. Cognitive radio: The cognitive radio, as its name implies, builds on Software-defined radio to carry a level of cognition or intelligence that permits decision-making and learned patterns of behavior. According to IEEE, the cognitive radio is a radio transmitter/receiver that is designed to intelligently detect whether a particular segment of the radio spectrum is currently in use and to jump into (or out of) the temporarily-unused spectrum very rapidly without interfering with the transmissions of other users.

Ths is old technology in cell sites

[Animats]

Cell sites have used software-defined radios for many years. Cell sites today have far fewer discrite radios than they have active channels. Here's a typical software defined cell site radio system.

This isn't all that new. It's just becoming cheap enough that it's worth doing for single-channel units.

Joint Tactical Radio System

[Anonymous Coward]

For some really interesting reading on software defined radios, do some research on the DoD's program called Joint Tacticle Radio System (JTRS). The goal is to replace ALL radio's in the DOD and other government agencies (think police, fire, ems, homeland defense, fema, etc as well as coalition countries) with a JTRS radio (in a few different form factors) using standard waveforms that are kept in a single library to ensure compatibility. The basic chasises will use pluggable modules for both "standard" waveforms as well as others that require either highpower, atypical freqs, as well as modules for encryption. the base system will be able to act as a gateway for compatible msg types on different wave forms allowing joining of systems that previously are not able to share traffic.

Then of course there is the Wideband Networking Waveform (WNW) which will also go on the JTRS radio. THAT is an interesting beast that is VERY ambitious. Not sure how much info is out there open source (even the unclass stuff) but whatever is out there should prove intersting. WNW is planned to be the SINGLE waveform for all DOD use after they phase out the others sans uber special ones like ULF for subs, etc. Think a completly networked battlefield for text, voice, video, and data on land sea and air, that can dynamicaly heal and reroute itself as nodes come on and off the network (for whatever reason be it "death" or loss of connectivity) all while keeping special forces hidden (but still connected), ground pounders situationally aware of where other friendlies are and where we know the enemy is via a networked display, instant access to survielence at the lowest levels and doing it all while under electronic attack ... as well as some other very nifty features.

To put it in perspective of how serious they are: NO ONE in the DOD is allowed to buy ANYTHING that isn't JTRS compliant without a waiver from the SecDef level ... and those are IMPOSSIBLE to get without a "if we don't have this now troops will die YESTERDAY" justification.

Re:Software radios a step towards real deregulatio

[mesocyclone]

I find it sad that the first post rated "Informative" is almost all political.

Can't we look at this technology without the technology government bashing and utopian (and ignorant) libertarian rants?

SDR is not a new technology, but it is rapidly becoming a good way to do things, as the hardware (digital and analog) to enable it is being designed and built.

Cell phone companies are (or will soon be) using SDR to much more efficiently handle their multichannel cell sites. Instead of having a radio per conversation, or a radio per channel, they can have one or a few radios containing very high speed DSP SDR code. This saves cost and has the obvious flexibility of field upgradeability.

GNU ( http://www.gnu.org/software/gnuradio/doc/exploring -gnuradio.html#software ) has had an SDR project going for quite a while (I do wish they would do APCO P-25 reception, since I don't have the time). Hams have been doing various forms of SDR also - for example, the very narrowband systems that use a PC to do the DSP for HF data communications.

Contrary to what some might think, SDR doesn't give magical powers to radios - the ability to operate on all frequencies at once. Radios have hardware filters in them for reasons that cannot be solved in software: to compensate for the non-linearities in the analog (or digital) software - which especially causes problems in high dynamic range situations. Radios may have to separate signals that differ in power by factors of 10^12 or more, which are relatively close in frequencies. Transmitters have to avoid emitting spurious signals at similar ratios to their output power.

More specifically, if you put two signals (assume sine waves for now) into a non-linear device, it is the equivalent of putting those time-domain functions into a polynomial of degree 2 or more. This means that those sine waves will be multiplied by each other and themselves (and a coefficient which you try to make as small as possible). The result is output at the sum and difference frequencies and the harmonics of the original signals. Non-linearity can crop up in surprising ways. The most common one seen in radio is receive and transmit amplifiers, which are *always* non-linear. In addition, parasitic devices (such as two wires touching each other somewhere nearby) can act as non-linear mixers, generating spurious signals. Anyone who has worked on systems at crowded radio sites knows the fun of tracking down "intermod" signals (which are the result of this process). SDR's do nothing to improve this situation. On the contrary, they may require wider bandwidth amplifiers, which increases the odds of spurious signals. Furthermore, the very process of sampling with non-infinite bit-width A/D's and D/A's is itself a non-linear process that generates mixing.

So SDR still has to deal with the issues at the antenna that analog radios deal with.

Where it gets cool is at the baseband - in other words, at the modulation=baseband level (or in the case of multi-channel receivers/transmitters, at an intermediate level). This is where you take the information you want to send/receive, and convert it into/from the RF representation of that information. A simple example is FM modulation (used in most older land mobile radios - police, fire, cell phones, ham repeaters, etc, and in TV and FM radio broadcast). Here the SDR will take the modulation (voice or music or whatever), and use it to generate the signal equivalent to having it quickly alter the frequency of a carrier wave. Depending on the system, it may literally output a sine wave modulated this way. In other systems, it may generate some intermediate representation that then goes to the radio.

But a far more interesting system might be a trunked narrow-band digital public service radio system (which US public safety organizations are converting to at FCC insistence). These systems are designed for improved flexibility (

Re:Software radios a step towards real deregulatio

[scotty1024]

In software radio utopia the bitch down the street would have a software radio that would automatically range the boyfriend and select a band that allowed for the most efficient and localized transmission. Which I sincerely doubt would be your precious highly regulated HF band.

In software radio utopia the radios would only invade low bit per second "buggy whip" HF bands when something like a category 5 Hurricane Zelda struck and knocked out the local high bandwidth high frequency ISP nodes.

And even then a UHF or VHF band would be more likely to reach a functioning ISP node. So don't get your HF panties in a bunch.

You assume I'd want to go WiFi

[scotty1024]

Here in Seattle I've often found my Verizon EVDO is faster than T-Mobile's Starbucks or Border's WiFi hot spots.

And what if I'm on the bus traveling down the street: 3G, WiFi, 2.5G, WiFi...

The decision to switch from 3G to WiFi will have to be made on more complex criteria than simply "Oh look WiFi!!"

Right now my Tablet PC can't even handle going "Hey Wifi!" reliably, although Mac's do it quite well.

And I can't even begin to picture how one would handle a TCP hand off with out using IPv6. RIght now Verizon and CIngular both suck at handing off seamlessly from 3G to 2.5G and back to 3G when running around in a bus on their own networks (where they have control over IP addresses' and routing).

I submit that these issues push things further out than you think to achieve your utopia.