DIY Passive Radar System Verifies ADS-B Transmissions | Hackaday

Blah2 Features

Using the blah2 software available at: https://github.com/30hours/blah2

• Support for the SDRplay RSPDuo, USRP (only tested on the B210) and HackRF.
• 2 channel processing for a reference and surveillance signal.
• Designed to be used with external RF source (for passive radar or active radar).
• Outputs delay-Doppler maps to a web front-end.
• Record raw IQ data by pressing spacebar on the web front-end.
• Saves delay-Doppler maps in a JSON array.

Using DAB (digital radio) as the transmitter, with a center frequency of 204.64 MHz and a bandwidth of 1.56 MHz. 

Running a CPI time of 500 ms and processing each CPI in around 600 ms. See the radar live online at: https://radar4.30hours.dev

setting up passive bistatic radar

Summary

The video describes a radar surveillance system setup. The main components include:

1. A 3-meter long Yagi antenna with seven directors, a driven element, and a reflector on the roof.
2. An ADS-B truth antenna at 1090 MHz and a reference antenna (telescopic monopole).
3. An antenna interface box connecting the roof antennas to the server rack.
4. A server rack with an SDRplay RSP Duo as the radar receiver, an RTL-SDR as the ADS-B truth receiver, and a HackRF as a spectrum watch.
5. A desktop PC acting as the radar processor, running software available on GitHub.

The software supports various software-defined radios and allows configuration of sampling rate, center frequency, and device type. The system displays a delay-Doppler map showing targets detected by the radar (orange dots) and the ADS-B truth data (green dots). Other displays include max hold, detections in delay and Doppler over time, spectrum reference, and timing information.

The transcript also discusses the detection of a helicopter (POL56) through its micro-Doppler signature and two aircraft (VH-418 and JST499) coming in to land at Adelaide Airport.

Video Transcript

Here is the video transcript:

On the roof, I have my radar surveillance antenna, which is a 3 m long yagi with seven directors, a driven element, and a reflector. Each of the elements are supported by two eyebolts connected to a base plate, which means I don't need to drill into the boom and can still move the elements back and forward as I need. The driven element is a dipole, which is separated by a 3D printed non-conductive base plate. There is also a nylon bolt on either side supporting this. The feed is COA cable, which has been separated into the inner and outer conductors. This connects to the common mode choke balun, which presents a high impedance to common mode currents and separates the antenna from the rest of the COA cable. The antenna is horizontally polarized, which means there is minimal interference from the vertical mast.

On the bottom here, I have the ADS-B truth antenna at 1090 MHz, and on the other side, I have my reference antenna, which is a telescopic monopole. Note there is also a small common mode choke here as well. The antennas on the roof connect down to the antenna interface box. On the bottom of this box, there's a 3D printed plate which has 16 N-type connectors. The eight at the front connect to antennas on the roof, and the eight at the back connect inside to the server rack.

Inside the box, we have our reference and surveillance antennas for the radar. These are loopbacks at the moment as there are no filters or amplifiers in the line. The third antenna is the ADS-B truth antenna, which runs at 1090 MHz. There's a lot of loss in the RG58 cable at this frequency, and I needed one of these SAW filters and low noise amplifiers to boost the signal to reasonable levels. The next two are just dummies. On the end, we have a 5-volt connector which comes from the server rack inside and this powers the ADS-B filter and LNA. These devices here are lightning discharge tubes. Here you can see we have a gas discharge tube which connects between the center conductor and the body of this connector.

The cables from the antenna interface box go up through this conduit and through to the server. The coax comes into the server room through to the back of the server rack. The SDRplay RSP Duo is used as the radar receiver, with the reference and surveillance channels shown here. The RTL-SDR is used as the ADS-B truth receiver, and the HackRF is used as a spectrum watch. A 5V power supply is used to the antenna interface box. All of the signals here are connected to a desktop on the other side of the rack.

Here's the front of the server rack with one of the desktop PCs being the radar processor. The code for the radar software is available on GitHub. At the moment, we support three different software radios, primarily the SDRplay RSP Duo, but we also have support for the USRP and two HackRFs chained together. The main thing is setting up the config file, where you can set things such as the sampling rate of 2 MHz, the center frequency for this example which is 2046.4 MHz, and the device we're using which is the RSP Duo.

On the delay-Doppler map, we have a target in the scene. Detections from the radar software are given by orange dots, whereas detections from the truth are given by the green dots. Here we have a flight, VH-418. If we go to the ADS-B truth display, we can see VH-418 is an aircraft coming in to land at Adelaide Airport. On the x-axis here, we have bistatic range, which is the distance from the receiver to target plus the transmitter to the target minus the transmitter to receiver. On the y-axis, we have bistatic Doppler, which is the same but for Doppler in the direction of the transmitter and receiver.

Looking at the other displays on the radar, we have a max hold display, which is useful to show some of the persistence of the targets. Essentially, we take the last 10 or 20 delay-Doppler maps and plot the maximum value for each cell. Interesting for this target here, we can see because it has sidebands above and below the main body return, that this is something with blades, and this is known as a micro-Doppler return. If we go back to the ADS-B display map, we can see POL56, which is a helicopter.

Looking at the other displays, we have detections in delay over time. This is useful for separating the false alarm detections from the true detections. Similar for detections in Doppler over time, once again you can see that micro-Doppler above and below the body return here. The spectrum reference shows our 2 MHz bandwidth and it shows that we have 1.5 MHz of signal, which is our digital radio. The timing display is useful to see which parts of the signal processing are taking the longest. Here we can clearly see the clutter filter and the ambiguity processing are the most demanding parts of the system.

The API data is shown here. Every time we refresh, we get a new delay-Doppler map, and this is the data that's used to plot. Similar thing for detection data. Back to the delay-Doppler map, we can see we have another target here, JST499. Going back to the ADS-B truth, another aircraft coming in to land.

DIY Passive Radar System Verifies ADS-B Transmissions | Hackaday

hackaday.com

---

Like most waves in the electromagnetic spectrum, radio waves tend to bounce off of various objects. This can be frustrating to anyone trying to use something like a GMRS or LoRa radio in a dense city, for example, but these reflections can also be exploited for productive use as well, most famously by radar. Radar has plenty of applications such as weather forecasting and various military uses. With some software-defined radio tools, it’s also possible to use radar for tracking aircraft in real-time at home like this DIY radar system.

Unlike active radar systems which use a specific radio source to look for reflections, this system is a passive radar system that uses radio waves already present in the environment to track objects. A reference antenna is used to listen to the target frequency, and in this installation, a nine-element Yagi antenna is configured to listen for reflections. The radio waves that each antenna hears are sent through a computer program that compares the two to identify the reflections of the reference radio signal heard by the Yagi.

Even though a system like this doesn’t include any high-powered active elements, it still takes a considerable chunk of computing resources and some skill to identify the data presented by the software. [Nathan] aka [30hours] gives a fairly thorough overview of the system which can even recognize helicopters from other types of aircraft, and also uses the ADS-B monitoring system as a sanity check. Radar can be used to monitor other vehicles as well, like this 24 GHz radar module found in some modern passenger vehicles.

Combining ADS-B with Passive Radar

The article "ADS-B information based transmitter localization in passive radar" by J. Yi, X. Wan, Y. Fu, and G. Fang proposes a novel method for localizing transmitters in passive radar systems when their locations are unknown in advance. This is important for target localization using time-difference-of-arrival (TDOA) techniques.

The proposed method combines automatic dependent surveillance-broadcast (ADS-B) information from aircraft with passive radar measurement information. The localization model is established, and the uniqueness and accuracy of the localization are discussed. Numerical analyses verify the feasibility of the method.

Key points:

1. Traditional passive radar systems assume known transmitter locations for TDOA localization of targets.
2. The proposed method localizes unknown transmitters using ADS-B information from aircraft and passive radar measurements.
3. The localization model is based on the bistatic range measurements forming hyperbolas with the aircraft positions and transmitter as focuses.
4. Single differencing is used to eliminate a constant factor related to the transmitter in the measurements.
5. Cramer-Rao bound (CRB) analysis shows that wider view angles of the aircraft trajectory with respect to the transmitter provide better positioning accuracy.
6. Numerical analyses demonstrate the feasibility of the method and the existence of good transmitter localization areas.

The proposed method improves the flexibility of passive radar systems in practical applications by requiring only the addition of a portable ADS-B receiver to the system.

"ADS-B information based transmitter localization in passive radar," 

J. Yi, X. Wan, Y. Fu and G. Fang, 

2014 XXXIth URSI General Assembly and Scientific Symposium (URSI GASS), Beijing, China, 2014, pp. 1-4, doi: 10.1109/URSIGASS.2014.6929185. 

Abstract: Transmitters' locations are usually assumed to be known in passive radars that exploit the third-party radio sources as illuminators of opportunity to detect targets of interest. However, there are such cases where transmitters' locations are unavailable in advance, causing difficulties for target localization. 

To address the problem, this paper investigates a novel transmitter localization method that combines the decoded automatic dependent surveillance-broadcast (ADS-B) information and passive radar measurement information. The localization model is established at the first, followed by the discussion of localization uniqueness and accuracy. The feasibility of the proposed method is further verified by numerical analyses.

keywords: {Passive radar;Radio transmitters;Receivers;Accuracy;Positron emission tomography;Aircraft},

URL: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6929185&isnumber=6928981

Published in: 2014 XXXIth URSI General Assembly and Scientific Symposium (URSI GASS)

Date of Conference: 16-23 August 2014

Date Added to IEEE Xplore: 20 October 2014

Electronic ISBN:978-1-4673-5225-3

DOI: 10.1109/URSIGASS.2014.6929185

Publisher: IEEE

Conference Location: Beijing, China

SECTION 1.

Introduction

Passive Radars, also known as passive coherent location, exploit the third-party radio sources as illuminators of opportunity to detect reflections from the environment and targets of interest. Passive radar is usually equipped with two type of receiving channels: reference and surveillance channels. The reference channel serves as a source of the original transmitted signal and the other one a source of the target echo signal. The noisy and multipath-polluted signal in the reference channel should be first purified as reference signal. Clutter rejection is executed in surveillance channel to cancel the clutters that may mask the target signals. Thereafter, by performing a 2-D cross correlation between the surveillance and reference signals, the echo can be detected by localizing the correlation peak in the range Doppler (RD) map. As no need for deployment of the dedicated transmitters, passive radars operate covertly, immune to anti-radiation missiles and active directional jamming inherently [1]–​[3].

Passive radar often acquires the measurements about bistatic range and Doppler shift, direction of arrival (DOA) also included in some cases. Time-difference-of-arrival (TDOA) localization [4] is used for target localization. The prerequisite of TDOA localization is to know the Transmitter s' locations in advance. However, there are such cases where transmitters' locations are unavailable aforehand. It could occur when the system works in strange environment where the transmitter stations are not covered in our database or in a developing broadcast network where newly installed transmitter stations are also unknown, or the transmitter stations stay at a special region we cannot get access to. In this case, transmitter localization becomes a necessary step if we want to implement TDOA localization. To the best of the authors' knowledge, the transmitter localization problem has not been discussed in reported references. An in-depth study on this problem would improve the flexibility of passive radar in practical applications.

Different from the traditional passive ESM (Electronic Support Measurement) traker (PET) [5] where the source is usually moving, the transmitter localization herein usually faces static transmitter. Thus the tracking method used in PET is not applicable here. Besides, spectrum management monitoring vehicle also play the role of radio station localization. However, the time cost is usually high. In this paper we propose an alternative automatic dependent surveillance-broadcast (ADS-B) information based transmitter localization approach for aerial target detection passive radar, where ADS-B is a cooperative surveillance technology for tracking aircraft in which the aircraft determines its own position via GNSS and periodically broadcasts this via a radio frequency. The condition is that there exists aerial target equipped with ADS-B device, which is easy to reach since ADS-B devices are common in civil aviation aircrafts. Thus the proposed method works quickly and only an ADS-B receiver is required to be installed on the passive radar system.

SECTION 2.

System Overview

To interpret the problem discussed in this paper, we first take a look at the typical systematic architecture of passive radar and traditional PET systems. Passive radar system is sketched in Fig. 1(a). As mentioned in Section 1, both reference channel and surveillance channel are configured in the receiver. The dotted line denotes the motion trajectory of the target. The transmitter, target and receiver construct a bistatic triangle (i.e. OAB). The TDOA obtained in the 2-D cross correlation step denotes the difference between transmitter-target-receiver path and transmitter-receiver path (baseline), i.e. bistatic range. Assuming the transmitter-to-receiver geometry is known, the TDOA indicates that the target is located in the ellipse/ellipsoid with transmitter and receiver as the focuses and with the sum of bistatic range and baseline as the major axis. Target location is determined by multiple such ellipses/ellipsoids that stem from multiple transmitter-receiver tuple. This is the basic principle of TDOA localization.

In our topic, we want to find out the position of the transmitter to utilize the TDOA localization. The similar systems that fulfil source localization are PET system and spectrum management monitoring vehicle. As shown in Fig. 1(b), PET system involves just radio source and receiver, no third-party illuminator included. It tracks moving radio source with information of DOA, Doppler shift, and so on. The receiver needs to do maneuvering motion to accomplish observability under some cases. By means of DOA and/or received signal strength, the spectrum management monitoring vehicle approaches the radio source gradually, arriving at the target eventually, as shown in Fig. 1(c). Nevertheless, source localization with spectrum management monitoring vehicle usually requires high time cost.

It is noticed that the passive radar itself is a receiving devise. The method with less additional equipment and low-cost operation would possess good practicality and flexibility. Observing the passive radar system, in the bistatic triangle, if transmitter and receiver are known, then we can localize the target with TDOA information. An alternative idea is that we can localize the transmitter with TDOA information if the target's position is known. Moreover, we notice that modern civil aircraft carries ADS-B system which broadcasts its position and velocity information. Thus, if there are civil aircrafts in the passive radar coverage, ADS-B information based transmitter localization would be a promising approach.

Fig. 1.

System sketch map of passive radar, PET and spectrum management monitoring. (a) Passive radar. (b) Pet. (c) Spectrum management monitoring.

Show All

SECTION 3.

Localization Model and Method

According to the bistatic geometry illustrated in Fig.1 (a), the bistatic range measured by passive radar can be expressed as

r=AB+OB−OA+e,(1)

View Source where e denotes measurement noise. With the aid of ADS-B information, then point B is known. Thus AB - OA can be obtained, indicating that point A is located on the hyperbola with 0 and B as focuses and with AB - OA as the real axis. The common intersection of multiple such hyperbolas is the position of transmitter.

Multiple such hyperbolas require multiple bistatic range measurements. The multiple bistatic range measurements can be provided by multiple targets, or by measurement sequence of single target, or the combination of them. When multiple targets are involved, it should be assured that the measurements of the multiple targets stem from the same transmitter. It is intuitive in multiple frequency network (MFN) where each transmitter works on a unique frequency. However, it is difficult to achieve in single frequency network (SFN) where the measurement to transmitter association is ambiguous since all the transmitters in the SFN broadcast the same signal at the same frequency simultaneously. Thus measurement sequence of single target will be the first choice in the SFN scenario because the measurement sequence of single target can be associated in the RD domain.

Moreover, in the SFN scenario, the extracted reference signal usually corresponds to a certain transmitter that maybe different from the considered one. Thus there is a constant factor difference between the estimated bistatic range and AB+OB−OA in practice. To include more cases, we extend model (1) to a general one that is

rk=ABk+OBk−OA+u+ek,k=1,2,⋯,N,(2)

View Source where u denotes the constant factor that is related to the transmitter. N denotes the number of measurements. The measurement noise ek is assumed to be independent with each other and follows Gaussian distribution with zero mean and variance σ2k.

TDOA localization method cannot apply to model (2) directly as there is the constant factor. Single differencing can solve this problem. By choosing rk as a reference, subtracting it from all other N−1 bistatic ranges gives

rk−rN=ABk−ABN+OBk−OBN+ek−eN,k=1,2,⋯,N−1.(3)

View Source

It indicates that point A is located on the hyperbola with Bk and BN as focuses and with ABk−ABN as the real axis. If N≥M+2, where M is the coordinate dimension of A, and Bk(k=1,2,⋯,N) are non-collinear, it can be concluded that the equation set (3) is solvable and the solution is unique. If N≥M+2 and Bk(k=1,2,⋯,N are collinear, the equation set (3) has two symmetrical solutions with the line, even unsolvable if A is collinear with Bk. If the solution is unique, it can be solved with least square (LS) or maximum likelihood (ML) method.

In addition, point B is assumed to be known in accordance with ADS-B information in the above model and method. In practice, this assumption needs to resolve the association between radar measurement and ADS-B information, namely determining the radar measurement corresponding to which target in ADS-B information. A feasible way is to exploit the DOA measurement which is a parameter independent to transmitter's position.

Finally, the practicality of the proposed transmitter localization method depends on its positioning accuracy. Cramer-Rao bound (CRB) is derived here to represent the positioning accuracy. Record m=[r1−OBk,⋯,rN−OBk]T, where superscript “T” denotes transpose operation, r,rR and sk denotes the position of the transmitter, receiver and Bk, respectively. The Fisher information matrix with respect to r and u conforms to

I(r,u)=∑k=1N1σ2kakaTk,(4)

View Source where ak=[(r−rR)/||r−rR||−(r−sk)/||r−sk||−1]. Then CRB can be obtained from the diagonal elements of I−1(r,u).

Interestingly, if we draw the vectors r−rR||r−rR||−r−sk||r−sk||(k=1,2,⋯,N) together, it will manifest as Fig. 2 where 2-dimensional case is considered. It shows that the support domain of r−rR||r−rR||−r−sk||r−sk||(k=1,2,⋯,N) increases with the increase of the view angle of target trajectory with respect to transmitter. According to the eigenvalue decomposition characteristics of I(r,u), bigger support domain indicates lower CRB, namely higher positioning accuracy. Thus, wider view angle provides better positioning accuracy under the condition of same number of measurement. In addition, more independent measurements will improve the positioning accuracy under the same view angle case.

SECTION 4.

Numerical Analyses

To verify the proposed method more intuitively, we do several numerical analyses in this section. The scenario of the numerical analyses is sketched in Fig. 3. There are two aircraft trajectories that represent two different cases. Both the two trajectories are π/2 arc with common focus at A2. 300 aircraft measurements are uniformly distributed on each trajectory. The accuracy ofbistatic range measurement is assumed to be 30m, i.e. σ1=⋯=σN=30mA1,A2 and A3 are three selected points for CRB comparison. 2-dimensional case for transmitter is considered.

Fig. 2.

Geometric interpretation of the CRB.

Show All

Fig. 3.

The scenario of the numerical analyses.

Show All

The CRBs of the selected points are presented in Table 1. The CRBs of point A2 in both cases are the same. The CRB of point A in the small arc case is less than the big arc case since there is wider view angle in the small arc case. The CRB of point A3 can be explained with the same way. All these are consisted with the theoretical CRB analysis in Section 3. The contours of the CRB in the whole 80km×80km area are shown in Fig. 4. In practice, if the positioning accuracy is on the same level of bistatic range measurement, the target localization performance degradation caused by non-ideal transmitters' position can be ignored. Specifically, in the considered case, the area that the root squared CRB is less than 30m could be thought as good transmitter localization area. It can be observed in Fig. 4 that the good transmitter localization area is relatively large. More importantly, the good transmitter localization area can be predicted according to the geometry. Thus the proposed ADS-B information based transmitter localization approach is feasible.

Fig. 4.

Contours of the CRB. The three columns correspond to CRB(x)−−−−−−−√,CRB(y)−−−−−−−√ and CRB(x)+CRB(y)−−−−−−−−−−−−−−−√

Show All

Table I CRBs of the selected points

SECTION 5.

Conclusion

This paper has discussed a novel transmitter localization method by combining the ADS-B information and passive radar measurements. Its feasibility and performance has been demonstrated through theoretical and numerical analyses. As only a portable ADS-B receiver is needed to be installed on the passive radar, it will greatly improve the flexibility of passive radar in practical applications. Although the proposed method has been verified by real-life data, the result with real-life data is not presented in the paper because of the limited paper length.

ACKNOWLEDGMENTS

This work was supported by national scientific fund committee of China under grant (61331012, 61371197, U1333106, 61271400, 41106156).