Sitemap Contact Events New Home

Database of research projects FEDRA


Research actions



Research and applications > Projectendatabanken > Database of research projects FEDRA

An Innovation Project on Multistatic Opportunistic SAR (MUSAR)

Research project SR/00/120 (Research action SR)

Persons :

Description :

This project deals with passive synthetic aperture radar. Passive radars use transmitters of opportunity as signal source and are necessarily bistatic. Passivity has many advantages:

- Built-in redundancy since any transmitter (subject to central frequency and bandwidth requirements) can be used as source;
- By exploiting several transmitters of opportunity, it is feasible to increase the revisit rate of a particular area;
- Lower cost than an active system since there is no need for a transmitter. Moreover since no signal is transmitted, there is no need for acquiring a license or obey transmit power regulations.

Passive SAR is to be seen as an enabler of other applications since, due to its low cost, it puts SAR imaging at the reach of lower budgets.
The most straightforward passive SAR system consists in using an existing SAR sensor such as ENVISAT as transmitter of opportunity and having a static (non moving) receiver. The direct and the backscattered signals are received and processed at the receiver to produce recurrent images of a given site at each sensor pass. Starting from this scheme, more complex systems are possible. Non-SAR transmitters of opportunity can be considered since any signal is usable provided some minimum requirements on the power density, the central frequency and the autocorrelation function are met. In particular, GPS and Galileo signals could be used (although much weaker than ENVISAT signals) and would provide a permanent imaging capability. Configurations involving static transmitters (typically numerical TV (DVB-T)) could also be used together with mobile receivers (on UAV, on ground vehicles).


The aim of the project is the development of a “proof of concept” passive SAR imaging system with a static receiver. The research activities are naturally divided in two parallel branches with strong interaction between them:

- Development of the aperture synthesis algorithms, on one hand using a beam-forming approach (“exact” algorithm) to prove feasibility and on the other hand using Fourier-transform based methods (“fast” algorithm) in order to potentially optimize execution speed. The optimized Fourier-transform based algorithms will clearly be one of the innovative points of this proposal. This also implies the modelling of the received signals, in order to have ideal data sets to test the synthesis algorithms.
- Hardware development and associated signal processing, where the hardware acquires the signal while the extraction of the reference signal and of the backscattered signal is performed in software. This software-based separation, whereas other passive SAR implementations perform signal separation in hardware using dedicated antennas, is another innovative point of the proposal.


In order to ease the general development, we consider the ENVISAT spacecraft as transmitter of opportunity. This has two major advantages:
1) the signal that is transmitted is very easy to “recognize” (it is a chirped signal) and
2) the power density on ground is relatively high. There is however one drawback: the number of passes. The cycle length of ENVISAT is about one month, which means that acquisitions with an optimal geometry will only be possible once or twice per month. This is somewhat to be nuanced if acquisitions in sub-optimal geometries are considered. This explains the necessity to separate the development and the test of the signal acquisition part from the development of the image synthesis part. It should be noted that the ERS-2 spacecraft operates in the same band and could also be considered, which would double the number of available passes.

The hardware acquisition device will consist in a four-channel receiver with the associate antenna array. Down conversion to intermediate frequency is performed in hardware and digital down conversion is used to convert the received signal to baseband. This is a similar setup as the one we used in [RD1,RD2]. Classically, the separation of the received signal in a direct path signal and the scattered signal is partially performed in hardware, using distinct antennas, each with its own orientation [RD3], for each signal. This has the practical drawback that the “direct signal” antenna needs to be pointed towards the source, and that source might not always be at the same location. We will consider real-aperture spatial beamforming in a similar way as we did in [RD1], where a static setup was considered. Software signal separation is clearly innovative and has a major impact on the actual usability of the system. The drawback of this approach is that, besides the required array calibration step, a large dynamic is required since the direct signal is typically much more powerful than the scattered signal. Our experience [RD1,RD2] is that a good separation of the direct signal from the scattered signal is necessary in order not to attenuate the scattered signal when the contribution from the direct signal is subtracted. If software separation is not sufficient, the direct signal can be synthesized, which is particularly easy in the case of chirped signals since the few required parameters can be estimated from the received (contaminated) direct signal.

The image synthesis will be performed using two different methods. In the first method, a direct application of a matched filter is considered and can be seen as performing beamforming over the synthetic aperture. This can be considered as a very good approximation to the inverse problem [4]. This method is expected to be very computing intensive. The main aim of this method however is to validate the received signals and associated processing. Indeed, this method is relatively simple and implementation errors will be relatively easy to detect. If necessary, a refinement of this method could be considered, as discussed in [RD4]. The second image synthesis method we will consider is an evolution of high performance SAR synthesis algorithms with a focus on high processing speed. The extension of high performance SAR synthesis algorithms to bistatic SAR is clearly innovative. The image synthesis will first be performed on synthetic (simulated) signals as explained above, in order not to delay algorithm synthesis development and in order to develop those algorithms in ideal conditions. These simulations will also give a first rough idea of the achievable performances (range, …). While in the practical realization, we will concentrate on ENVISAT signals, the simulation will also consider transmitters of opportunity that emit pseudo-noise signals, such as GPS or Galilleo satellites.

Once an image is obtained, its quality will be assessed. It should be stressed that the features (corner reflexion, …) found in the usual (monostatic) SAR images will probably not appear in a bistatic image. Actually, very little is known about bistatic scattering. Absolute quality measures shall be attempted. In any case, Equivalent Number of Looks (ENL) shall be estimated. If a point scatterer is detected, Integrated SideLobe Ratio (ISLR) and Peak-to-SideLobe Ratio (PSLR) will be derived and compared to the values predicted by simulation. The quality of the image will not only depend on the quality of the processing but also on the quality of the hardware. We will attempt to separate the influence of the hardware (signal to noise ratio, issues of signal dynamic) from that of the image synthesis algorithm itself. To that end, we will perform a comparison of the quality of the images obtained using the synthetic beamformer approach with the image obtained using the high performance SAR algorithm.


The questions we would want to answer are fourfold. What are the necessary hardware modifications, if any, that can be done to improve performance? Is software separation of the direct and scattered signal achievable and what is the price to pay (in terms of image quality or imaging range) for that convenience? How complex is a fast SAR processing algorithm and what is the effect of the required approximations? And finally, what are the performances in terms of imaging range, actual resolution, … The answers to these questions will serve to precise the scope of follow-on projects.

Documentation :

About this website


© 2017 Science Policy PPS