|Design and Analysis of Compressive Sensing Techniques for Radar and ESM Applications
|Sensors & Electronics Technology|
Adaptive Signal Processing, Beamforming, Compressive Sensing, Convex Optimization, Data Reduction, ESM, High Resolution Radar, Imaging Radar, Inverse Synthetic Aperture Radar (ISAR), Performance Assessment, Robustness, Sparse Signal Representation., STAP, Synthetic Aperture Radar (SAR), TWI
Compressive sensing (CS) is a novel acquisition and processing theory that enables reconstruction of sparse signals from a set of non-adaptive measurements sampled at a much lower rate than required by the Nyquist-Shannon sampling theorem as pertaining to the full signal bandwidth. In particular, compressive sensing exploits the fact that the information bandwidth of the signal is much smaller than the full signal bandwidth. Rigorous results have shown that signal reconstruction using CS can be achieved by convex optimization. A tutorial article by Stephen Boyd in the May 2010 issue of the IEEE Signal Processing Magazine illustrates the potential of these techniques for regularizing a range of ill posed inverse problems.
In the NATO framework, the SET-213 Specialist Meeting on Compressed Sensing for Radar/SAR and EO/IR imaging that took place in Tallinn, Estonia, in May 2014 provided an opportunity for specialists in this research field to come together, and discuss about the state-of-the-art and way forward to bring CS radar and EO/IR systems to the next level. The applications discussed during the Specialist Meeting emphasized several advantages of using CS based systems. For example, significant hardware reductions can be obtained in ESM receivers by using CS; unambiguous signal recovery from incomplete measurements (filling missing data) is possible in interleaved radar modes (SAR/GMTI, multi function radars) and sparse arrays; high resolution imaging via MIMO radars or multi-pass (e.g. tomography) can be achieved with significantly less data and/or hardware. In 2014-2015 the SET ET-093 was created to further investigate the application of CS to radar and ESM and it received broad consensus amongst several NATO nations.
Most of the work carried out so far in CS applied to radar has been focused on demonstrating, either theoretically or with simple experiments (mostly based on simulated data), that CS can be successfully applied to radar imaging, detection, direction finding and classification problems. Furthermore, signal model simplifications have often been used to be able to deal with the mathematical aspects that may undermine the solution robustness when applied to real world scenarios. Therefore, there are still many open questions and practical issues that need to be addressed in order to bring CS radar and ESM from a theoretical exercise into an operational system. In particular, performance analysis of such algorithms and sampling schemes in non-ideal conditions is necessary. For example, in operational scenarios the noise might not be Gaussian, and the performance in the presence of correlated clutter and jammers should also be investigated. Furthermore, the robustness of CS based techniques against errors such as imperfect knowledge of the receiver impulse response, off-grid targets, and covariance matrix estimation for STAP/beamforming with limited training data need to be carefully analysed.
The benefits and challenges associated with CS techniques should be indentified and recommendations should be given to NATO regarding their use in existing and future operational systems. Cooperation amongst several NATO countries would provide a framework for sharing data and algorithms, which are necessary for investigating the performance and robustness of these methods.
The main objectives of the RTG are
• Create a common shared algorithms and data repository;
• Define metrics and procedures for performance and trade off analysis with specific attention to robustness, resiliency, and reliability;
• Assess the performance and robustness of CS based system architectures and algorithms for radar and ESM applications/scenarios;
• Identify and quantify benefits and challenges in the transition to operational CS based radar and ESM systems;
Furthermore, information will be exchanged with other NATO groups working in related areas, such as the SET-RTG-090 on “Computational Imaging and Compressive Sensing for EO/IR Systems”, and possibly jointly organize a workshop or specialist meeting.
Four major applications will be investigated, namely:
1. high resolution imaging (SAR, ISAR and through the wall imaging (TWI));
2. SAR GMTI and STAP;
3. Direction of Arrival (DOA) estimation, tomography and beamforming;
For each application the advantages and challenges of using CS-based architectures and algorithms will be investigated and its performance and operational benefits evaluated using simulated and, when available, real data.
The analysis will include:
1. Definition of performance indicators such as false alarm and detection probabilities, classification rate, minimum detectable velocity, accuracy, probability of resolution, contrast and resolution for imaging systems, memory requirements, computational load and convergence for implementation of CS recovery algorithms;
2. Robustness to modelling errors, clutter and noise distributions, parameter selection;
3. Operational benefits and challenges of CS for considered applications;
4. Trade off between performance and computational load;
5. Feasibility and ease of integration in current systems and computational architectures;
6. Comparison with other existing high resolution methods (e.g. MUSIC and Burg’s algorithm).