Defcon 19: build your own radar system

Posted on Monday, November 14th, 2011 in DIY, RF, talks by the machinegeek

At the Defcon 19 (2011) conference Michael Scarito demonstrated how to construct your own radar system. The project is based on a design by Dr. Greg Charvat from MIT and uses synthetic aperture techniques to generate a two- or three-dimensional image. (You can read Dr. Charvat’s research paper here.)
The hardware operates in the 2.4 GHz ISM band (shared with WiFi.)

For a further detailed explanation of this technology, see the dissertation of Manh Hung V. Le, Dimitris Saragas and Nathan Webb from the Worcester Polytechnic Institute here.

A free MIT course on this topic is available through their Opencourseware presentation “Build a Small Radar System Capable of Sensing Range, Doppler, and Synthetic Aperture Radar Imaging.”

This entry was posted on Monday, November 14th, 2011 at 4:34 pm and is filed under DIY, RF, talks. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

8 Responses to “Defcon 19: build your own radar system”

  1. Drone says:

    1. Early-on in the presentation, the important concept of chirp in CW-FM radar is not explained well at all. Also, it is not until the very end that a Doppler related question from the audience evokes the mentioning of up-down chirps.

    For a brief explanation of chirps in radar see the section on “Frequency Modulation” here:

    2. Why is there no IQ (quadrature) demodulation? In applications like this quadrature demodulation simplifies almost all DSP tasks. Hardware quadrature demodulation is preferential for RF down conversion (over digtal delay quadrature) and is cheap and easy to accomplish in this application. The demodulated bandwidth of interest is relatively small compared to the carrier frequency, so quadrature dividers would be used to split the local oscillators, then image reject mixers would be employed for down conversion. All are relatively inexpensive COTS parts (, and with a little more design effort, they can also be built for next to nothing as microstrip elements directly on the PCB (decent quality FR4 board will work fairly well at 2.4GHz, see for the mixer diodes.).

    3. Why build a complex 250kHz data acquisition system? Truly excellent yet affordable high dynamic range 24-bit 192kHz sound cards are available today (even as USB dongles) – and they’re stereo, so accepting quadrature input is a no-brainer.

    4. The use of foul language in these public presentations is insulting; not only to the audience (me), but to the character of the speaker. If I were the Author’s Adviser at MIT; I’d be doing a bit of advising about this.

    Now, where did I put my Prozac?…

    • Jack says:


      Foul language? There was nothing worse than normal workplace conversational language. Or are you one of those prissy self-censoring dinosaurs? Let me know when you’re having your retirement party, I’ll send a bowdlerized card.

  2. rsdio says:

    Thanks, Drone. I haven’t even watched yet, and already I’ve learned a few interesting things.

  3. rsdio says:

    3) There are more differences between a 24/192 audio A/D and a purpose-built 250 kHz A/D than you seem to be allowing for. Cheap 24/192 sound cards are great for audio, because the 8 bits of quantization noise that they have can be easily ignored by the human hearing system. But radar processing probably isn’t going to be too happy with that same noise when it comes to imaging from faint signals. A 250 kHz A/D is also cheaper than you might think, and it can be a precision part that is vastly better than an audio A/D. I think that the speaker did a good job of explaining that the 250 kHz sample rate is needed to generate a smooth ramp. A 44.1 kHz ramp would have serious stepping, and those steps would cause FM problems. 192 kHz might be fast enough, but 24-bit audio DAC chips have plenty of noise shaping, where it’s fairly cheap to get a DAC that is less than 24-bit but avoids the sigma-delta modulation that creates all that noise. Sometimes, other DAC technologies are way more suited for non-audio purposes.

  4. SOI Sentinel says:

    I’ve been reading and researching this design over the last few weeks, got some thoughts for you, Drone. Of note, I am an interested hobbyist, not an RF engineer nor an ARRL RF guru.

    The entire front end was designed by the professor for the summer course. It’s about as simple as you can get, which was the point. I’m not certain IQ would help in FMCW, most notes I find apply that to pulse doppler. Are they actually doing any quadrature on this? I thought it was all based upon frequency analysis via FFT and rejecting the Q component (I peeked at the Matlab files).

    I can’t say I agree with either Drone or rsdio on the A/D conversion. I’d have to look into noise floors and ENOB at frequency. I’m not exactly convinced that he needed an integrator, either, on the front end. More research still ahead.

    As for the presentation, I did enjoy it, it cleared up a few points I had questions on, but it was more like an end of class “this is what I learned” presentation, not something that extended the knowledge of his professor.

    I’d love to see an RF front end board made, as $250+ of connectorized parts is out of my budget, but students in a class are not expected to have MMIC waveguide design experience, either. I’d also like to see it pushed up the frequency band (5.8GHz) if possible, just to get away from most interference. 24GHz is just asking too much… maybe. This original design appears to be based upon the low cost building penetrating RADAR/SAR setup that Dr Chavat has been working on for years, so I don’t think they ever even considered going up band.

  5. SOI Sentinel says:

    Thought about this a bit m ore. Drone, are you saying most of this could be implemented with the integrated RF front ends I see some vendors selling for Wifi/Wimax crowd?

  6. magetoo says:


    Truly excellent yet affordable high dynamic range 24-bit 192kHz sound cards are available today (even as USB dongles)

    Watch the talk (around 23 minutes), he *did* use a sound card in the first version. Just a bit later he explains how the data acquisition board makes things easier.

    The use of foul language […]

    You have got to be kidding. (And never go to a hacker conference.)

    SOI Sentinel:

    but it was more like an end of class “this is what I learned” presentation

    Yeah, it’s a talk given at Def Con. Think “telling your buddies about this cool thing you did”, not “presenting a project”.

  7. david says:

    just interested in how to build my own radar if you could send me more info that would be great

Leave a Reply

Notify me of followup comments via e-mail. You can also subscribe without commenting.

Recent Comments

  • Siddharth: Grate
  • jeanmarc78: hello
  • KH: So I guess, don't mind us dinosaurs too much. Someone like me will get triggered by some of the things Arduino folks do. There's just...
  • Max: I wouldn't begin to argue educational aspects, that is not my field and various approaches are just too... arguable. I do have to say however...
  • samuel: have a nice week everyone