Data Repository of Passive Radar Measurements
Data sets listed in the VEGA repository can be used only for research purposes.
Any commercial use of these data sets is forbidden.
This database is open, contributions to the repository are welcome.
If you are interested in using the database, please send me a private request in which you briefly summarize your research goals and the planed way you would use the data sets.
Click on the Title of the data set to open the detailed description of the measurement !
The VEGA data repository stores passive radar measurement. Each data set in the repository has unique name with the following naming convention:
VEGA<M/S><Date in yyyymmdd format><Hardware id>U<unit id>C<Test configuration ID>S<Surveillance ID><IoO type>
e.g.:VEGAM20190101HR2U0C0S0FM. The first identifier (<M/S>) declares whether the data set is originated from measurement or simulation. Each measurement has metadata fields which describes the particular circumstances of the measurement or simulation environment. These fields are the following:
The primary stored data in the data sets are the IQ samples. These samples are recorded in rather large data chunks, called coherent processing blocks. Each of these coherent processing data blocks are packed into frames by the data acquisition system, named as IQ data frames. An IQ data frame is built up from a header and a payload. Each binary file which has '.iqf' extension in the 'iq' folder stores one IQ data frame. The user can import these data frames, interpret the header, extract the raw data and utilize it for his own research purposes. Documentation and reference scripts to import the IQ data frame are available bellow in this page.
Figure 1: Standard folder structure of a VEGA dataset
Beside this, there is suplementary information available where applicable. Figure 2 shows the standard folder structure. The results folder stores range-Doppler matrix and hit matrix pictures from reference processings. The root folder of the K-th reference processing is named as PK. The parameters of each of the reference processing is summarized in the 'dsp_params' file.
The target_information folder stores the real latitude and longitude coordinates of the target with timestamps if available. For aerial targets this information is obtained from FlightRadar24. For other cooperative targets, GPS tracker is used to obtain the same information. Other than this, the actual range and Doppler coordinates of the detected target is also available for all the coherent processing blocks in which valid detection can be found.
The following list summarize the available target information fields and their designation.
Target information fields:
List of the identified targets in the data set
Each target has the following descriptor fields:
Target ID : (Unique ID, start indexing from 0)
First time index : Index of the first coherent processing block in which the target can be detected (according to reference processing)
Last time index : Index of the last coherent processing block in which the target can be detected (according to reference processing)
Target denomination : (e.g.: Landing airplane/ moving card, closing train, receeding pedestrian,... )
Target type : (e.g.: A320)
Target RD trajectory : Range and Doppler values of the degected target
Real Target trajectory : (Ideally this should be a list of lat,long positions)
Documentation of the IQ Data Frame can be found here:
Version: 1.1, Updated: 2020 01 27
Example python code to convert the IQ Data Frames to MATLAB compatible file
can be found in the following GIT repository:
You can cite the used data set in the following format:
Tamas Pető (2019), Vega: Data Repository of Passive Radar Measurements, <DATA SET ID>, [Data set]. http://tamaspeto.com/vega
L. Bendjama, T. Laroussi and A. Mezache, "Statistical Analysis and New Modeling of Real Clutter Signal in FM Radio-based Passive Radars," 2019 6th International Conference on Image and Signal Processing and their Applications (ISPA), Mostaganem, Algeria, 2019, pp. 1-6.