Universal Radio Grabber: the USRP

Nethemas the Great writes "The Universal Software Radio Peripheral or USRP created by Matt Ettus and Eric Blossom gives a new perspective on the radio spectrum, as in just about all of it from DC to 2.9Ghz. With the right software and daughterboards, their USRPs can capture FM, read GPS, decode HDTV, transmit over emergency bands, track peoples movement via their mobile phones, and much, much more. With prices starting at just $550 this new toy is accessible by most anyone."

Comments

Uh, prices don't begin at $550

[drinkypoo]

Just the motherboard is $550. You will need at least one daughterboard to actually do anything. The cheapest ones (2-200MHz transmitter, 2-300MHz receiver, 30MHz transmitter, 30MHZ receiver) are $75 each. In order to just transmit, you will need to spend at least $625, unless you are a member of "TAPR, AMSAT, SARA, or SETI League" in which case you get $25 off the motherboard.

Interestingly, though the sales page lists "extra" power supply, usb cable, and standoff sets, nowhere on the sales page does it actually say that the unit comes with any of these things. If you're going to run a business, run it right.

Re:Ouch $550

[davidbro]

Especially since it appears that the $550 gets you the motherboard, without any radio modules. The $550 will get you something that you need to spend more money on until it is functional. I think this is just a case of the journalist not really digging into it more than superficially, but the guy the reporter was talking to should have also pointed out how much a minimally configured system would cost.

At a minimum you will need the motherboard, a radio module, some cable (which isn't cheap, especially for doing higher frequency work), and a useful antenna (those tiny ones they advertise on the website will be fine for higher frequencies, but if you want to do anything else, you are looking at an external antenna and more cable).

However, this is a very cool project. A lot of good will come from this work. But $550 is not the starting price. The starting price is higher.

Nothing new here...

[Jizzbug]

It has been on the market since Nov. 2004.

http://www.comsec.com/wiki?UsrpProgress

Re:How is this legal?

[prichardson]

From their FAQ... http://www.ettus.com/faq.html

Are there any license requirement for the transmit or transecive daughterboards?

The USRP is sold as test equipment, which has no licensing requirements. If you choose to use your USRP and daughterboards to transmit using an antenna, it is your responsibility to make sure that you are in compliance with all laws for the country, frequency, and power levels in which the device is used.

Depends on the country

[Sycraft-fu]

In the US it's not legal to have a device that listens in on certian bands, such as cell bands and military frequencies, and other than a few speicifc bands, you need a license for any transmitter. So the transmission components are almost certianly illegal in the US, at least to use. The reciever components, it depends on the range, and if the have holes where they should for given disallowed frequencies.

Now this applies to the US one, other countries do not necessairly have an FCC equivilant that regulates such things.

Re:Depends on the country

[mindriot]

This product seems to receive the entire spectrum by default.

No. The USRP motherboard is capable of handling anything from DC to 2.9 GHz, but you need the matching daughterboards for specific ranges. Daughterboards include:

• BasicRX, 0.1-300 MHz receive
• BasicTX, 0.1-200 MHz transmit
• LFRX, DC-30 MHz receive
• LFTX, DC-30 MHz transmit
• TVRX, 50-860 MHz receive
• DBSRX, 800-2400 MHz receive
• RFX400, 400-500 MHz Transceiver
• RFX900, 800-1000 MHz Transceiver
• RFX1200, 1150-1400 MHz Transceiver
• RFX1800, 1500-2100 MHz Transceiver
• RFX2400, 2250-2900 MHz Transceiver

Also, you obviously need to have the matching antenna to actually receive something useful in a given frequency range.

Now, whether or not receiving particular frequencies is allowed or not will obviously depend on the FCC and similar regulatory organizations (in most, if not all countries, for instance, receiving police radio frquencies is illegal). Maybe the FCC regulation you mentioned is taking things a bit too far... cell phone standards like GSM are encrypted anyway (unless, of course, you go for a man in the middle attack).

As to your FCC quote, I suppose the question is whether being able to buy another daughterboard/antenna means it can be "readily altered to receive such frequencies." With respect to transmitting, the FAQ states that since it's sold as test equipment, you don't need a license. I wonder if the "test equipment" status supersedes that FCC statement as well?

Re:The real question

[zippthorne]

Transmitters must be licensed. Even your 49 mhz walkie-talkie is licensed. It's not licensed to you specifically, it's type accepted, so the manufacturer can sell lots of them. If you were to modify it to transmit on another frequency, you would have an unlicensed transmitter and therefore subject to prosecution if you actually used it.

It makes sense to license transmitters. The EM spectrum of useful radio frequencies has finite bandwidth, and we must have some plan for use so that the most people can get the most benefit out of it. This includes astronomers, hobbyiests, emergency services, cell-phone users, television studios, and many more. Licensing solves the traffic jam problem.

It makes much less sense to license receivers. The radiation is there, passing through people, even. Frankly, I don't understand why anyone would think that I don't have the right to intercept any signal which passes through my personal space and process it however i please.

But that seems to be the case. Recievers capable of recieving cell-phone frequencies may not be sold. I am unsure of the legality of modifying or building your own equipment for that purpose, but I am sure the cell-phone companies have lobbied hard to make that illegal as well. As a longtime desirerer of encrypted cell-phones, it has frustrated me that they want to transmit "in the clear" and just make it a crime to recieve, especially as equipment from before there were cellphones exists that has no hardware blocks on those frequencies whatsoever. Fortunately, CDMA forces at least a rudimentary level of quasi-encryption.

I hear hype...

[SuperBanana]

"Here," he explains, "I'm grabbing FM." "All of it?" I ask. "All of it," he says. I'm suddenly glad the soundcard isn't working.

Not quite- in order to fit the swath of FM radio into that USB2 pipe, it isn't sampling it in any great detail. If you tried to decode one station, it'd most likely sound like a tin can, unless you sampled a narrower slice of the FM band. So don't get too excited. Claiming the motherboard or these devices are "universal" is extremely misleading. You buy modules that transmit or receive on different bands. They're usually pretty wide in frequency spectrum, but they also generally aren't anywhere near as good as dedicated receivers for those bands, and they're not "universal."

Claims of being able to receive GPS are also misleading- you'd be able to decode individual satellites and perhaps obtain a fix within a mile or so, but getting accuracy anywhere near what a $100 handheld GPS unit can do, would require incredible timing accuracy that board just doesn't have. Remember...GPS works by timing how far radio waves w/time signals take to travel...down to about 10 feet in some cases. Think hard about what kind of timing accuracy and precision that requires.

Re:I hear hype...

[lowen]

I have two of these personally. At PARI we have four of them. They work. And work well, for radio astronomy.

As to capturing the entire FM band at one fell swoop, I know for a fact that the USRP and a good USB 2.0 High-Speed host can sustain 32MB/s transfers. This becomes an actual sampling rate of 8MS/s in quadrature, which means a full 8MHz band can be sampled at 12 bit precision. The FM band is 107.9-87.9=10MHz wide. At 12 bits, no, you can't get the whole band in. However, the USRP can go 16MS/s at 8 bits (again, in quadrature, which effectively doubles the sample rate), and consume 32MB/s across the USB. Since FM (frequency modulation) doesn't require large dynamic range in terms of bit depth, it is possible that you could get nearly full fidelity audio out of all FM channels simultaneously: but you would need one big honking PC to demodulate in real-time.

As I am a licensed Amateur, I can use this as a transmitter, in the bands and with the modulations to which my license class is allowed. I have the 400-500MHz transciever board; I am of course limited to the 70cm Ham band for transmission, and I of course honor that. It works quite well.

For radio astronomy, I have the DBS_RX board, and it directly tunes several radio astronomy bands, including the Hydrogen line at 1.42GHz. It works quite well for both continuum and spectrum studies, although I still have some bugs (with considerable help for the GNUradio project and other programmers) to work out.

Re:I hear hype...

[lenhap]

You obviously read the article, but did you think to read any details on the actual device? The baseboard/motherboard has a ADC that can capture 10 million samples per second at 12 bits per sample. So doing simple math and ignoring protocol overhead to transmit all 10 million samples would require 12bits per smaple * 10 million samples per second = 120 million bits per second. USB2 has theoretical bandwidth of 480 million bits per second, so the rough back of the envelope calculations would suggest that the full 10 million samples the ADC can capture could be transmitted to the computer over USB2.

So if we assume that the all of the data can get to the computer, could the device grab all of the FM in such a format that it could be "decoded" into normal FM quality audio? Short answer, yes. The daughter cards for the baseboard/motherboard convert the signal down to an IF (intermediate frequency) within the range of the ADC. If you really want to know how IF and all that stuff works, look up FM radio on wikipedia.

What really annoys me is how you try (key word is "try") to explain that this device cannot do GPS. You do NOT need accurate timing to do GPS. Time is part of the GPS solution, so you only need a simple realitively accurate clock. The $100 handheld GPS units don't have anything more accurate than the clock in your pc, which this device would have access to (the clock in your pc, that is). In fact GPS is often used to provide timing for applications like NTP servers. Again you would need one of the daughter cards to convert the GPS signals down to an IF. The actual GPS signal (C/A-Code) is transmitted in the L1 band (1575.42 MHz) which when converted down to an IF could be handled by the ADC in the device. From there you would only need to aquire 4 satelites to get a simple PVT solution (position, velocity, and time). And, FYI, GPS in certain applications and situations can give accuracy to within cm range (mm range if using differential GPS and post processing which this device could do).

So before posting as if you are an expert, look up some stuff on what you are writing...or at least explain that you aren't positive on how everything works but you don't think it could do what it claims. And yes, I actually work doing military GPS for a company and have a BS in EE with a concentration in communications (so I should hopefully know what I am talking about).

Similar package for less than $600 complete

[ps_inkling]

The ICOM PCR1500 (Japanese) already receives everything from DC to 1.3GHz (minus analog cell frequencies, unless you're a government user). No additional modules required, and uses USB and fairly open software controls.

Or, for even cheaper ($350), Ten-Tec's RX-320D, with digital radio. Everything from DC to 30MHz (shortwave).

I've never used any of them, your milage may vary, etc.

So what?

[Slithe]

IP Adresses can be changed, and MAC addresses can be spoofed. If you are TRULY paranoid, connect to a random Access Point with a spoofed MAC address and talk using an encrypted VOIP connection. Simple, easy, and cheap (you can buy a laptop, microphone, and wifi card for less than the cost of the USRP motherboard.

Re:How is this legal?

[Bloke down the pub]

The USRP is sold as test equipment [...] If you choose to use your USRP and daughterboards to transmit using an antenna, it is your responsibility to make sure that you are in compliance with all laws

Reminds me of a story I heard: during prohibition, you could buy a health drink that was basically grape juice concentrate. The instructions said something like "Do *not* dilute and certainly don't add yeast. If accidental yeast contamination occurs, don't even think about leaving it in a warmish place for roughly two weeks".

"the right daughterboards"

[Erandir]

Be careful of that seemingly innocuous qualification: "with the right software and daughterboards"... both imply serious limitations to the technology.

Firstly, the "right" software: Even with a reasonably fast processor (say 3 GHz) today, you are typically only be able to process, at most, a few million samples per second -- especially if you are performing complicated modulation/demodulation, coding/decoding, filtering and protocol processing. Each sample may require substantial computation, and that limits the number of samples you can process per second. That, in its turn, affects the bandwidth that a processor can address (i.e. how wide a part of the radio spectrum you can "see" at any one time).

Secondly, the "right" daughterboards: To be able to address a wide bandwidth, we require digital-to-analog and analog-to-digital converters with high sampling rates. These are limited by the state of the art in signal conversion technology -- typically a couple of million samples per second if we want a reasonable number of bits per sample (at a reasonable price). Push it beyond that, and we have to be happy with fewer bits per sample (may 10 or 8 bits). This introduces noisiness to the signals being transmitted or received, degrading the fidelity of the software-defined radio.

Also, a daugterboard usually has some form of signal translation hardware ("mixers") to translate the low-frequency signals that computers can generate to and from the higher parts of the radio spectrum. Although broadband mixers are available, they need tunable oscillators (reference frequencies), and these tend to be limited to narrower parts of the spectrum. Also, analogue filters, amplifiers and antennas (which all form part of a typical software radio front-end), usually are limited to specific ranges of the radio spectrum.

In short, software radio daughterboards tend to be fairly application-specific (or at least spectrum-specific). We can do a lot of things in software, but a "universal" software radio needs a lot of hardware swapping. I think that makes it a bit less "universal". It might also push the cost of a truly multi-purpose system quite a bit beyond $550.

But I'm glad to see this technology receiving such mainstream attention, and I applaud the efforts of the designers. I just think that TFA (and the post) could maybe be a bit less sensasionalist.

And yes, IAASDRE.

G-J

Re:decoding HDTV?

[TheRaven64]

The Slashdot summary was monumentally bad. My first thought on reading it was 'this sounds like the kind of thing you could use with GNU Radio." Clicking on TFA, I discovered GNU Radio was in the article title. Strange how Slashdot, usually GNU-obsessed, would miss that out.

GNU Radio is a pretty amazing piece of software. I attended a talk about it at Linux '05, and was amazed by the capabilities. When they say they are decoding HDTV, they mean that they are doing it in software. All of it. Not just decoding the MPEG-2 streams, but everything this side of the analogue to digital convertor. They are not running it through a decoder box and grabbing it from a FireWire connection, they are capturing the radio signals, converting them into digital signals in hardware and then doing everything else in software.

The basic architecture of GNU Radio is a filter API. Individual filters are written in C++ for performance and then they can be joined together and controlled with Python, making the barrier to entry very low for anyone who wants to tinker with it. Don't be fooled into thinking you need an expensive receiver like the one in TFA to play with it either, it will accept input from a large number of ADCs, including sound cards. You can use it to apply transformations to any digital waveforms.

You can use it to implement something like 802.11 entirely in software, generate telephone dialling tones on your sound card, modulate your voice to sound like a Dalek, decode HDTV signals, or a huge range of other things. It turns your PC into a hugely powerful programmable DSP.

The hardware in TFA is just icing on the cake. As I recall, the specs for a slightly simpler model are available from the GNU Radio site, so you can build one yourself if you have (a lot) more time than money.

timing BETWEEN SIGNALS, nanosecond range

[SuperBanana]

Sorry, GPS location requirements don't rely on the timing on the board, all of the timing and position is derived from the received signal. You need to be able to receive 3 or more satellites for a fix

I don't think you understand how GPS works. Simplifying- a GPS receiver looks at when signals with the same timestamp arrive, and deduces how far it is from each satellite from that. If a signal from Satellite A saying "hey, it's 12:01:05 right NOW arrives a second after a similar signal from Satellite B, then the receiever knows that it is 1 light-second further away from Satellite A than B (this is a gross exaggeration of the scale of time involved.) With 3-4 satellites, you get a position fix.

Modern receivers can track 12-20 satellites at once and get accuracy down to 10 feet or so. There are two things the receiver must do which are timing-related:

1)Figure out what time it -really- is, so it can set an internal chronometer, so it can know the exact distance it is from satellites, versus relative distances

2)Record as exactly as possible when each satellite's particular timestamp came in

Both require -staggering- accuracy that a PC, or your USRP board, are incapable of providing. Clock skew considered perfectly acceptable in a PC is considered monumentally inaccurate in a GPS receiver...and the timing resolution isn't anywhere near good enough either. You're talking about comparing timing in LIGHT FEET, and light takes 1/299,792,458th of a second to travel a meter. It's about one NANOSECOND a foot, so you need resolution exceeding 10nS.

You've got to do a lot of signal processing to ignore spurious signals, as GPS signals love to bounce off some things, and get absorbed readily by others. You've got to have an incredibly low noise, highly sensitive receiver, as GPS is readily absorbed by just about anything, and that includes trees.

The current state of the art is SiRF's SiRF-3 chipset; I've got a Garmin handheld with one, and I can get a 30 foot position lock inside my house, under treecover. I can get a 10 foot lock if I'm outside with enough satellites in view and a WAAS differential signal. I'd -really- like to see you try to beat that.

Re:timing BETWEEN SIGNALS, nanosecond range

[Anonymous Coward]

I don't think you understand how GPS works. Simplifying- a GPS receiver looks at when signals with the same timestamp arrive, and deduces how far it is from each satellite from that. If a signal from Satellite A saying "hey, it's 12:01:05 right NOW arrives a second after a similar signal from Satellite B, then the receiever knows that it is 1 light-second further away from Satellite A than B (this is a gross exaggeration of the scale of time involved.) With 3-4 satellites, you get a position fix.

You are the one who has no clue here. Very few GPS systems work that way, because it requires an atomic clock. Instead, they solve for 4 variables. 3 for position and 1 for time. Any page explaining GPS will tell you this. Look it up yourself.

Re:timing BETWEEN SIGNALS, nanosecond range

[David Bengtson]

Sorry, I know exactly how GPS works, I'm working on a GPS receiver board for Gnu Radio (www.keystoneradio.com).

You don't need absolute timestamps on each sample to get distance. The timing is encoded in the PN sequence in the transmission from each satellite. Also included in the transmission from each satellite is the satellite orbit information and the exact time that the GPS system uses. First, you receive 1 satellite, and then set your GPS clock to the system clock. Once your handheld GPS is set to the system clock, you can figure out how far you are away from the satellite by determining time of flight from the satellite to the handheld (You know the PN sequence, and you know the time. You can figure out the time that you received the signal, and then you can determine the distance)

Once you know the distance from 3 satellites, then you know you are in 1 of 2 locations. If you further assume that you are reasonably close to the earth's surface, then you knock off one of the possible locations. More satellites will give you a better estimation by giving you more position estimates to average out.

As far as the signal processing, the GPS signal comes in to the antenna pretty close to the thermal noise floor, and so it's a bit tough to receive. Because it's a direct sequence spread spectrum system, you get 43 dB of processing gain from the de-correllators which helps significantly. Multi-path isn't really to much of an issue because the signal is primarly from overhead. Absorbtion is more of an issue, and is something that Garmin/SiFR etc spend a lot of time dealing with.

As far as beating the current state of the art in GPS receive algorith's with a half dozen guys working in their basement, probably not. As an educational thing, it's pretty useful.

Dave

Re:The real question

[drinkypoo]

So it should be legal for me to use a night-vision scope to look into my neighbor's bedroom window at night?

But it is legal. Anything you can see from your property is fair game to look at, at least in the USA. Probably not legal to record it; almost certainly not legal to distribute if you do.

It's totally legal to look at naked people in their own house if you can see them without trespassing. If you choose not to cover your windows you give up your reasonable expectation of privacy.

Amusingly you're not legally trespassing until you have been told to leave, at least in California. Those "no trespassing" signs don't mean shit either, unless your property is completely encircled with fence and you have a gate which is locked.

Re:decoding HDTV?

[maxume]

The novelty of GNU Radio is that instead of just the HDTV decoder being software, the tuner is too. If the application(TV, fm radio, HDTV, etc) you want falls inside the frequency range your hardware can sample, you can build a software tuner. The novelty of this hardware is that the cheapest prior solution started in the several of thousands -- generalized a/d converters intended for scientific data sampling, like: http://www.atmel.com/dyn/products/devices.asp?fami ly_id=611

Re:The real question

[cduffy]

Those "no trespassing" signs don't mean shit either, unless your property is completely encircled with fence and you have a gate which is locked.

Not here in Texas, at least. At our last Neighborhood Watch meeting, the officer who was speaking said that those signs count as a command from the homeowner to leave, and allow him the ability to arrest folks who are acting suspicious on someone else's property (whereas otherwise, the best he could do would be asking them to leave).

Now, in terms of shooting folks who are trespassing, the sign may not be sufficient alone (I haven't taken the concealed-carry class yet, and so am quite unfamiliar with the applicable law) -- but in terms of making trespassing into an arrestable offense, it is indeed.