Thank you to Daniel Kaminski for writing in and sharing with us news that he has recently updated his SDRDue Passive Radar software for RTL-SDRs. The major update is that thanks to NVIDIA CUDA GPU processing, the ambiguity function can now be calculated extremely quickly, allowing for very high frame rates. Daniel writes:
Last time I was playing with my Passive Radar. I finally created an ambiguity library which is a really fast 70 frame/s analyzing a continuous string of data 2*1024*1024 bits per frame. This allowed me to record signals from slowly moving cars in real-time. I used a normal TV antenna without any modifications in one dongle mode. To support the library I created a Passive Radar program with all the parameters available for tuning. The code is open and available on GitHub. The movie is available on my website Passive radar | Web page od Daniel M. Kamiński (umcs.pl).
If you weren't aware of it, KrakenSDR is our RTL-SDR spinoff project and is a 5-channel coherent RTL-SDR that we have successfully crowdfunded for over on Crowd Supply. KrakenSDR is the successor to our previous 4-channel coherent product called the KerberosSDR. With a radio like KrakenSDR that is capable of coherence between channels, interesting applications like direction finding and passive radar become possible. You can also use it as five independent RTL-SDRs should you chose to.
We wanted to note that all units preordered through the Crowd Supply crowd funding campaign are now at the Crowd Supply / Mouser warehouse, and the majority have already been shipped out to customers!
Additional units for new purchasers are in a mixture of production and freighting and will be available for fulfillment as soon as we can. We are constrained by supply and production time, so if you're interested in a KrakenSDR, please get your order in so that you have an earlier place in the queue.
Other Recent KrakenSDR Updates
Wiki Manual: Our Wiki manual and guide is up at https://github.com/krakenrf/krakensdr_docs/wiki. It covers topics from what you need to get started, radio direction finding theory and background, antenna array setup, KrakenSDR Web-GUI software guide, Android App guide and a Passive Radar guide.
Install Scripts, VirtualBox Images, Docker: For general vehicle based direction finding, which is the most popular application, we recommend using our premade Raspberry Pi 4 image for easy almost plug and play setup. But to ease installation on other computing devices (especially as the Pi 4 stock is non-existent at the moment due to the supply chain crisis) we've now created an automatic Linux install script and a Virtual Box image which can be run on Windows or Linux host machines. Third parties have also released a Docker container. See this page on our Wiki for more information.
Customer Feedback: We've also had some great customer feedback so far with one user submitting examples of his success in locating transmitters like a 162 MHz NOAA weather station, and various fox hunt beacons.
With a 5-channel phase coherent RTL-SDR interesting applications like radio direction finding (RDF), passive radar and beam forming become possible. It can also be used as five separate RTL-SDRs for multichannel monitoring.
Like many other projects we have been severely delayed by COVID work restrictions and the effects it's having on the supply chain, and I'd like to thank everyone who is keen to get a hold of a KrakenSDR for their patience. But the ball is rolling faster now and we have finally received our latest KrakenSDR prototypes! Testing has been ongoing for the last few days, and apart from a few minor issues everything is working brilliantly. At this stage we are confident in the design and are making plans to begin the crowdfunding campaign soon.
Supply Chain Constraints
The first batch will unfortunately be limited to 1000 units maximum due to supply constraints and we expect this first batch to be ready 2-3 months after the campaign finishes. So if you are after a unit ASAP, please ensure you are on the CrowdSupply mailing list as we fully expect demand for the first batch to outstrip the supply.
But if you are willing to wait, batch 2 will be still be available at the campaign special price. we will have a second batch available for early preorder at a discount (sorry due to higher than expected shipping and skyrocketing component prices we can't discount the second batch at the moment). Please keep in mind that the second batch will be at least 6 months away due to the long supply chain resulting from the pandemic.
The next stages in hardware development will involve finalizing our custom milled aluminum enclosure, testing one last prototype, and beginning mass manufacturing when the crowd funding campaign is over.
Work on the software is ongoing, but the beta version of our new DAQ firmware and direction finding DSP software layer is stable and already available on the krakensdr GitHub at https://github.com/krakenrf. Everything resides in the development branches and there is full documentation on the code structure available in the Documentation folder. This code can also be used on the KerberosSDR by editing the configuration files to specify 4 receivers instead of 5.
By the time the units ship out we will have a ready to use SD card image for the Raspberry Pi 4 and a quickstart guide available.
We have also been working at improving the Android direction finding companion app. This app was made during the KerberosSDR release a couple of years ago, and is used to plot and log the direction finding bearings being generated by the Kerberos/KrakenSDR unit, combining it against the GPS and movement data generated by the Android phone. This Android phone + KrakenSDR combination results in a powerful multipath resistant radio direction finding tool, and once enough data has been collected (usually after a few minutes of driving) it is able to determine where the most likely transmitter location is.
The upgraded app makes use of the full 360 degrees of direction of arrival and multipath data that is generated by the KrakenSDR, resulting in a more accurate determination of the transmitter location, and a better understanding of the uncertainties. It also allows users to visualize multipath. There are also various bug fixes and improvements made overall. We are planning to transition this app into a paid app, but all KrakenSDR backers will receive a license for free and the older KerberosSDR app will remain free.
To work as a radio direction finder, KrakenSDR needs five antennas. If you plan to use them in a circular array, they need to be omnidirectional antennas such as whips or dipoles. So to go along with the KrakenSDR we will be selling an optional set of five magnetic whip antennas which can be mounted on for example, the roof of a car. (Please note the magwhips shown in the photo may differ slightly from the final ones sold).
We have also been working with Arrow Antennas in the USA, who are producing a KrakenSDR 5-element dipole array antenna which is great for use in fixed sites (for example on the roof of a house). The antenna will be sold by Arrow antennas (not by us), and the future link (not active yet) will be http://www.arrowantennas.com/arrowii/kraken.html. We expect them to generate this page within the next few days. This antenna has been used in all our fixed site experiments as you can see in some of the YouTube videos, and works very well. (The image below show a prototype, we're told the final version may look slightly different.)
DAQ & Direction of Arrival (DOA / Radio Direction Finding) :
Work on the DAQ and DSP software is coming along well and this is mostly complete and runs stable on a Raspberry Pi 4. There are just now bug fixes and minor features being added. Intermittent 'bursty' signal handing is already working, but we are working on improving it's sensitivity to weak bursty narrowband CW signals which can still be problematic to detect. The Android app is also currently being field tested.
Work on new passive radar software is also ongoing and we expect to have something ready for experimentation and with quickstart guides before shipping. At the moment it is also still possible to use the older KerberosSDR software for passive radar, but we believe the new DAQ core software will run things much smoother. The goal for the new software is to not only plot a range-doppler map, but to combine it with direction finding and be able to plot radar detections on a map. This feature may require operation on a device faster than the Raspberry Pi 4, such as GPU based device like a NVIDIA Jetson.
Beam Forming, Interferometry:
One application we think the KrakenSDR would be great with is amateur radio astronomy via interferometry. The ability to combine multiple small hydrogen line dishes spread out over several meters of area should result in much greater radio imaging resolution, without needing to deal with a single huge dish. It may also allow for electrically steering a beam without needing to rotate the dishes.
Advanced Direction Finding + Advanced Log Management:
At the moment networked direction finding (direction finding via multiple fixed or mobile sites spread out around a city or area) is possible via the third party RDF Mapper software, but we aim to create our own advanced platform in the near future. The goal is to have software that will automatically log and alert when a signal of interest appears. For some examples we can see this being used to help coastguard locate distressed marine pleasurecraft that typically do not have AIS via their VHF radios, locate emergency beacons, for animal/wildlife/asset tracking, and monitoring for illegal/interference transmissions.
At this stage the core DAQ+DSP software will also be updated to support monitoring multiple simultaneous channels within the available 2.56 MHz bandwidth, and with a scanning and beacon ID detection feature.
Research into field applications:
One example we hope to test is the operation of KrakenSDR on a drone. With great line of sight from up in the sky, localizing a transmitter should be fast. Another example could be actually visualizing signals like light via augmented reality.
Some of our previous KerberosSDR and KrakenSDR posts might also be of interest.
If you weren't aware KerberosSDR is our 4-channel phase coherent capable RTL-SDR unit that we previously crowdfunded back in 2018. With a 4-channel phase coherent RTL-SDR interesting applications like radio direction finding (RDF), passive radar and beam forming become possible. It can also be used as 4 separate RTL-SDRs for multichannel monitoring.
In previous posts we've shown some interesting experiments performed with the KerberosSDR. For example:
We note that V2 of our KerberosSDR demo software is also on the way but a little delayed. We are aiming to release a beta around the end of the year, or early next year at the latest. The new software will have better handling of bursty intermittent signals, and paves the way for new developments coming in 2021 such as combined passive radar direction finding.
Passive Radar works by using already existing powerful transmitters such as those for TV/FM. A receiver listens for these signals being reflected off of objects like aircraft and vehicles, and compares the reflection with a signal received directly from the transmitter. From this information a doppler (speed) vs range graph of detected objects can be calculated and displayed.
By measuring the path an object travels across the range-doppler display some interesting information about the objects movement can be obtained. However, the display can be noisy, with the reflected object often coming in and out of view on the display. In order to track an object across the range-doppler display in the face of these uncertainties Max uses a Kalman filter to obtain smoothed results. A Kalman filter is an algorithm which combines actual data with predicted data, with the weighting depending on measurement confidence. The result is shown in the video below. A smooth and accurate track of an aircraft can be seen.
Max notes that in the future he'll be working on tracking multiple aircraft detected by the passive radar, and also incorporating direction finding data in his results in order to get cartesian coordinates which could be plotted on a map.
We note that Max's GNU Radio code should be compatible with our KerberosSDR unit, which already has the clock sharing hack built in to the hardware.
If you've been following KerberosSDR development (our US$149 4 channel coherent RTL-SDR), then you'll know that one interesting experiment that you can set up with it is a passive radar. Passive radar makes use of already exiting strong transmitters that broadcast signals such as FM, DAB and HDTV.
With one directional antenna pointing towards the transmitter, and one pointing in the general direction of moving objects like aircraft, it's possible to detect the transmitted signal being reflected off the aircraft's body.From the time delay and doppler shift detected in the reflected signal, a simple distance/speed plot showing the aircraft in motion can be created. This previous post shows an example of what information you could potentially collect in a range/speed graph over time. In the past we've also used passive radar to detect vehicles and measure how much traffic is in a neighbourhood.
However, with two antennas we can only get the detected object's range and distance information. If we use four antennas (one pointing towards the transmitter, and three pointing in the direction of objects), it is possible to use beam forming techniques combined to obtain an estimated map coordinate of the object. This is possible as we then we have distance information available from the passive radar algorithm, and bearing information available from the beam forming algorithm.
Tamas Peto who wrote our open source KerberosSDR code has been working on some new upcoming features for the KerberosSDR software, and beamformed direction finding of passive radar is one of them. We note that to be clear this software is not yet released, and we still expect there to be several months before it is ready. At the moment all data was processed manually offline after collecting data with a KerberosSDR as part of this early test.
The image below shows an example of a recent measurement made from an aircraft. The red tracks show the actual ADS-B GPS coordinates of the aircraft, and the black line indicates the positional data measured from a DAB signal reflecting off the aircraft body. The orange line to the east indicates the main lobe of the three beam formed directional antennas, and the lines to the west indicate transmit towers.
The measured trajectory is only about 1-2 km off the actual one. Tamas notes that the position offset may be because at the moment altitude is not measured yet.
Other upcoming features that are planned for the KebrerosSDR code include being able to use direction finding on short bursty signals, improvements to networked direction finding and beamforming which may be useful for applications like radio astronomy and performance improvements.
KerberosSDR can be purchased from the Othernet store or Hacker Warehouse, and every purchase helps us fund development of more interesting features like passive radar beamforming!
Over on his blog, Max Manning has posted about his senior year design project which was an RTL-SDR based passive radar system that he created with his project partner Derek Capone. Max's writeup explains what passive radar is, and how the theory works in a very easy to understand way, utilizing graphs and short animations to help with the understanding. The rest of the post then goes into some deeper math, which is also fully explained.
Passive Radar works by using already existing powerful transmitters such as those for TV/FM. A receiver listens for these signals being reflected off of objects like aircraft and vehicles, and compares the reflection with a signal received directly from the transmitter. From this information a speed/range graph of detected objects can be calculated
For hardware, the team used two RTL-SDR dongles with the local oscillators connected together. A standard dipole is used as the reference antenna, and a 5-element Yagi is used as the surveillance antenna.
Max's post is a great read for those trying to understand how to do passive radar with a KerberosSDR which is our 4x coherent input RTL-SDR unit available from the Othernet store or Hacker warehouse. Being a radio capable of coherency, it is useful for applications like passive radar and direction finding.
Their code is all open source and available on GitHub. We note that their code should also work with KerberosSDR with only either zero to minor modifications required. However, for the KerberosSDR we also have our own passive radar code available which might be a little easier to setup via the GUI.
Recently we've been testing a simple peak hold for the KerberosSDR passive radar display. This results in some nice graphs that show aircraft and vehicle activity over time.
Passive radar works by using already existing transmitters such as those for HDTV and listening for reflections that bounce off of RF reflective objects. With a two antenna setup, it is possible to generate a bistatic range/doppler speed graph of reflected objects.
With the reference Yagi antenna pointed towards a 600 MHz DVB-T tower, and the surveillance antenna pointed to an airport we were able to obtain the graph below. The top two large traces show aircraft heading towards our station, whereas the bottom traces show aircraft leaving the airport. Also visible are multiple blips with smaller doppler speeds, and these correspond to vehicles.
The code on the KerberosSDR git will be updated in a few days time. We are also working on a more comprehensive passive radar tutorial that will try to explain concepts like processing gain, bistatic ranges and other important tips for getting good passive radar results. At the same time we're also working on improving direction finding ease of use by prototyping antenna switches for calibration, and working on getting 4-channel beamformed passive radar working which will allow us to plot passive radar returns on a real map.