KrakenSDR: Passive Radar Demonstration

KrakenSDR is a 5-tuner coherent software defined radio based on RTL-SDR. It is the successor to the KerberosSDR and will be crowdfunded on Crowd Supply in a couple of months time. Please sign up to the KrakenSDR Crowd Supply mailing list to be notified as soon as the campaign begins.

The KrakenSDR (prototype - enclosure may change slightly)

Passive Radar uses existing FM, TV or mobile phone transmitters. The signal from these transmitters reflects off objects such as road vehicles and aircraft. By using two antennas on two receive channels, and an algorithm to compare the reflected signal against a clean reference copy of the actual signal, we can achieve a radar like display of bi-static range vs doppler speed.

In this test KrakenSDR is used as a two antenna passive radar system. The reference antenna points towards a horizontally polarized 620 MHz DVB-T transmitter, and the surveillance antenna points towards an Airport.

Passive Radar setup with two TV Yagis

Reflections of aircraft and road vehicles can be seen on the map as red dots/trails. Notice how we can also determine the overall neighborhood activity of road vehicles as we pointed out in a previous KerberosSDR post.

Of note is that we've placed the surveillance antenna in a vertically polarized configuration. With passive radar you want to keep the reference signal out of the surveillance channel, as ideally the surveillance channel only receives the reflections. Using the surveillance antenna in vertical polarization achieves 20dB attenuation of the horizontally polarized DVB-T signal. The reflections are assumed to be randomly polarized, so the vertically polarized antenna should pick up the reflection just the same no matter what polarization is used. This scheme woks especially well in our setup as the angle between the reference transmitter and target reflected objects is small.

This test uses the older KerberosSDR code (slightly modified to allow for trails), however new passive radar code is being worked on for the new KrakenSDR code base which will be released later this year. We expect the new code to also be able to make use of GPU accelerated CUDA hardware, such as the NVIDIA Jetson. This will allow for a much faster update rate and/or more processing gain.

The new KrakenSDR code will also try to make use of the additional three unused channels. With these extra channels we should be able to add a direction finding array that will help to plot on a map the actual location and elevation of the reflections.

KrakenSDR Passive Radar Demonstration 1


  1. Thom

    I also am looking at a boat application. A sailboat so slow moving. I see an issue with passive on the move because you are changing direction and dont have a direct point to the transmitter unless you built in a tracking device. I understand WIfi signal range is minimal, however what if we used a mono pulse in the 25-27 mghz range (Citizens Band Radio) in the legal 5 watts or say even up to 100 wats with a linear amplifier?

    • Seasalt

      The reason I chose a 2.4ghz pulse ie Microwave was in my mind the seawater around the boat would absorb the signal and any object on the ocean would reflect the signal.

      I thought that would sort of be a double win and make a cheap radar idea work better.

  2. snn47

    Will the USB-C design of the Kraken allow full bandwidth of each RTL-SDR without bits dropped?
    What is the bandwidth of the RTL-SDR achieved in the Kraken design compared to a RTL-SDR with USB2.0 out?

    • admin

      Using a USB-C connector doesn’t change anything to do with the data transfer, so the same maximum of 2.56 MSPS per unit still applies. It’s just a much more stable and sturdy connector.

  3. snn47

    1. I found during measurements, that the above made claim, that you achieve 20 dB of cross polar isolation between a horizontally transmit and a vertical polarized receive antenna within direct Radio Line Of Sight (RLOS) is in reality much lower. Driving around within 5 km direct RLOS and of a horizontally polarized VHF signal, the terrain was without any trees or bushes, just a mix of gras, concrete or asphalt and only a few meter of change ground height above mean sea level, the cross polarization isolation to be more in the range of 6 to 8 dB of, when measured with a commercial Tectronix spectrum analyzer and a tune whip on a bus.
    2. In varrying terrain with bushes and trees, and/or densely built areas, multipath usually reduces the cross polar isolation further.. For this reason DVB-T may also be transmitted in addition to horizontal polarization also with elliptical or circular polarization.
    3. While for horizontally polarization 100 % of the radiated power is radiated horizontally polarized, for circular polarization (50% of the power is radiated with vertical and 50 % with horizontal polariazition) while for elliptical polarization the <50% are radiated using vertical polarization. If circular polarization is used the cross polar isolation is at best 3 dB!
    4. You also need to keep in mind that the reflected signal strength and polarization of reflected signals vary and depend on the radar cross section and properties of a reflecting object (e.g. form and size of the reflecting object).

    • admin

      I think with a Yagi you are much more likely to achieve good cross polarization attenuation compared to with a whip. In the above tests the effect of horizontal vs vertical is significant. I actually do have another DVB-T transmitter nearby which transmits vertically, so I also need to be careful not to point the surveillance Yagi towards that tower. And yes you’re right, the polarization does very much depend on the object and radar cross section, but it seems that cars and aircraft do transfer from horizontal->vertical polarization quite well in my setup.

      Even if the antennas are kept in the same polarization you can get good results, but you will need the angle between the reference and surveillance Yagis to be larger. At least 90deg is good.

      Passive radar is very much a project that needs some trial and error with the antenna setup to get the best results.

  4. Seasalt

    What is approximate price?
    Would it work on moving boat with its own transmit source ie 2.4ghz wifi pulse etc?

    • Billy

      They answered the price being missing in another thread, I do not think that anything has changed since then:

      Nope. At the heart is five R820T2 tuners (which can typically tune between about ~24MHz to ~1766 MHz). There may also a direct sampling mode to go lower in frequency (no idea if that is available in the KrakenSDR), but to go higher you would need at a minimum external anti-alias filters and probably additional external gain for each antenna. Or use five external down converters with a common clock.

    • admin

      Price of Kerberos was typically anywhere from $149 to $299 depending on stock levels and component availability. Kraken will most likely be within that range too.

      And yeah as Billy said 2.4 GHz isn’t really easily possible with RTL-SDR hardware. You wouldn’t really get much usable range with 2.4 GHz WiFi anyway, unlikely enough to see another boat unless it was right up against it. The researchers who’ve experimented with 2.4 GHz WiFi typically did so with the targets only one wall away.

Post a comment

You may use the following HTML:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>