Let’s start off with proof. Below is an animation of a measurement of airplanes and meteors I made using a radar system that I built with a few simple easily available pieces of hardware: two $8 RTL software defined radio dongles that I bought on eBay, and two log-periodic antennas. And get this, the radar system you’re going to build works by listening for existing transmissions that bounce off the targets being measured!
I wrote about this in a very brief blog posting a few years ago. It was mainly intended as a zany little side story for our radio telescope blog, but it ended up raising a lot of interest. Because this has been a topic that keeps attracting inquiries, I’m going to explain how I did the experiment in more detail.
It will take a few posts to show how to build a radar capable of performing these types of measurements. This first part is the overview. In later postings I will go through more detailed block diagrams of the different parts of a passive radar system, provide example data, and give some Python scripts that can be used to perform passive radar signal processing. I’ll also go through strategies to determine that everything is working as expected. All of this may sound like a lot of effort, but don’t worry, making a passive radar isn’t too complicated.
Let’s get started!
A passive radar is a special type of radar doesn’t require you to have a transmitter. You rely on a radio transmitter of opportunity provided by somebody else to illuminate radar targets. This can be your local radio or television station broadcasting with up to several megawatts of power. The advantages compared with a normal radar are numerous:
- You don’t need a transmitter, which is a good thing because high power radio transmitters are large, expensive, power hungry, and contain dangerous parts..
- Being a pirate radar operator can sound exciting, but you really don’t want to transmit without an FCC license. Passive radar allows you to operate a radar with kilowatt or even megawatt class effective radiated power, without braking any laws or regulations.
- A passive radar system is inherently multi-static. A single station can multiple transmitters that can be in different locations. This allows the three dimensional trajectories of radar targets to be estimated. There are many radar targets that have illumination and viewing angle sensitive radar cross-sections. A good example is a specular meteor trail echo, which only has significant radar cross-section only when the illumination angle viewing angle combination is a specular reflection. With a multi-static system there is a greater chance that such a radar target is seen, as there are multiple different simultaneous illumination directions occurring at any given time.
- There are many radio transmitters out there: Television, FM radio, and AM radio are the most obvious ones, but cell phone towers and even satellites can be used for passive radar. These transmitters cover a wide range of frequencies, which is really useful for radio remote sensing, as radio propagation characteristics and scattering properties of different media and radar targets can be highly frequency dependent. For example, if a radar target doesn’t have a large enough radar cross-section on one frequency, there is a good chance that another frequency will work better.
While most radars systems we know nowadays have a dedicated transmitter, the idea of passive radar goes way back. One of the first radar experiments could be classified as a passive radar. In 1935, Sir Robert Watson-Watt and his colleague Arnold Wilkins performed the first aircraft tracking radar experiment by detecting echoes from a Heyford bomber aircraft illuminated by a BBC shortwave wave radio transmission. The figure on the right shows the two dipole antennas and the mobile laboratory used to perform the measurement. This seminal work was the basis of the Chain Home early warning radar system that went on to save numerous lives during the Second World War.
There are a large number of non-military applications for passive radars. I’m interested in the use of passive radar for geophysical and astronomical radio remote sensing. An example of a successful passive radar system used for geophysical remote sensing is the Manastash Ridge Passive Radar developed at the University of Washington, which was used to study ionospheric irregularities produced during geomagnetic storms. In the last 15 years the technology has become much more accessible, making it possible to perform these types of measurements with easily obtainable hardware.
The Manastash Ridge radar used FM radio for passive radar, which is probably the easiest to start with, because the bandwidths of individual stations are relatively modest. With FM radio, you can also typically observe not only ionospheric irregularities, but also meteor trails and movements of airplanes flying within a 100-600 km radius of yourself. A recent measurement that I performed during a geomagnetic storm is shown in the figure below. You can see large scale ionospheric irregularity structures that are moving across the field of view at 1000 meters per second associated with the aurora borealis north of the passive radar receiver.
The thing is, pretty much anyone can do these types of measurements themselves. That is if there was open source software to analyze the data. This would open the door for citizen scientists interested in a broad range of topics:
- Meteor surveys of meteor radiants. The CMOR radar found 12 new meteor showers just by looking at meteor radiants. While this was an active radar, the same could be done with a digital TV passive radar system. Radar allows you to observe a lot more meteors than you can optically.
- Mapping the locations of aurora borealis and the presence of strong ionospheric electric fields by observing echoes from ionospheric irregularities. This tell us about the interaction between the Solar wind and the plasma surrounding our planet.
- Measurements of atmospheric winds using Doppler measurements of meteor trail echoes. These are good tracers of the neutral wind at altitudes between 90 and 110 km.
If there were a network of passive radars around the world, there would be a great potential for new scientific discoveries to be made.
Example hardware
A surprisingly small amount of hardware is needed to build a simple passive radar:
- Personal computer (a dual core newer than five years old will do).
- Dual channel software defined radio.
- Two antennas.
An example of a receiver antenna pair used for FM radio passive radar is shown below. Both antennas are log-periodic antennas that are broad band in nature, but also have some degree of directionality. One of the antennas is pointing towards the FM radio transmitter and used to obtain a measurement of the waveform transmitted by the radar. The other is pointing towards the opposite direction and is used to measure echoes. All of the example measurements shown in this article were performed with these antennas.
The signals from these two antennas need to be recorded in a coherent manner. This means that we need two channels with samples that are aligned. This is typically achieved by using a common clock used as a reference for the downconversion stages and the analog to digital converters. The picture below shows two possible off-the-shelf devices that can be used for passive radar. On the left is a modifed dual RTLSDR R820T dongle system I hacked together for $16! On the right is a USRP N200 software defined radio with a dual channel TVRX2 tuner daughter board. A word of warning before rushing to implement the RTLSDR approach: there are subtle tweaks that need to be done in order to make everything work, involving aligning samples and aligning the center frequencies of the two data streams. The limited dynamic range also makes it much more difficult to get good measurements out of the system as the levels need to be carefully adjusted to get enough signal into the receiver while avoiding compression or clipping of the signal. But it can be done if you have enough patience! I’ll try to share what these tweaks are in the next post.
In order to get more fidelity out of a passive radar system, it is advisable to have a preamplifier and band-pass filter between the antenna and the software defined radio to reduce out of band noise contaminating the measurements and to increase signal levels so that they are more suitable for the digital receiver.
two RTLSDR R820T dongles modified to share a common oscillator.
Two alternative digital receivers that can be used for passive radar
a USRP N200 with a TVRX2 daughterboard capable of measuring two receiver channels
Signal Processing
In order to obtain passive radar echoes, we need to apply some signal processing magic to the digital signals recorded by the hardware. We first need to get rid of the strong direct path signal so that we can observe the weaker echoes. Robert Watson-Watt and Arnold Wilkins simply used an opposite phase signal to cancel out the direct path signal in their early Daventry experiment. Modern signal processing can do this much better, not just removing the direct path signal from the transmitter, but also reflections from mountain sides and other large scatterers that might mask weaker signals of interest. This is done by using a well known statistical signal processing technique called linear least-squares estimation to deconvolve the phase and amplitude of the direct path signal and the echoes from mountains and other strong non-moving scatterers (also called radar clutter).
After the strong direct path signal and clutter has been estimated, it can be subtracted from the measured signal and the signal processing to estimate the weaker echoes from airplanes or meteors can be performed. Because we know that the weaker targets that we are interested in have a significant Doppler shift and Doppler spread, we can’t make the same non-moving target assumption we did for the clutter. However, we can still make the assumption that the target has a scattering Doppler spectrum that doesn’t change over a long enough period of time to allow us to estimate the range-Doppler spectrum. This allows us to perform deconvolution on the autocorrelation function of the received echo. This is something that is called lag-profile inversion. While this sounds complicated, it really isn’t.
Once all of this is done, we can plot the results. If you see airplanes, your radar is working. While airplanes are boring radar observables, they are by far the easiest ones to see and they are there all of the time. Meteors are more transient, but also should be relatively easy to measure if you know what to look for. They appear as quick blips at near zero Doppler shifts.
Angle of arrival
There are also many different variations of FM radio passive radar that can be done. For example, by adding a third antenna to the system, one can perform angle of arrival direction finding by inspecting the phase difference (time of arrival) between two receiver antennas spaced apart from each other. The video below shows phase difference measured using two different receiver antennas. In this video, the phase difference between the two antennas is used to select the color of the pixel (hue) and the echo strength is used to color code the brightness of the pixel. Again, lots of airplanes and specular meteor echoes can be seen.
That’s all for this post. In upcoming posts I’ll go into more detail on how passive radar works in practice.
