Search

Space-Time Adaptive Processing (STAP) is a modern radar signal processing technique that leverages additional Degrees of Freedom (DoF) to cancel clutter from a background environment and produce detections of slow-moving targets. STAP is well-documented and understood; however, bistatic applications, or applications in which a radar transmitter and receiver are physically separated, present additional complications. This work explores techniques in Bistatic Space-Time Adaptive Processing (B-STAP) for Ground-Moving Target Indication (GMTI)---the detection of slow-moving surface targets through ground clutter. Due to the complexity and availability of B-STAP data, the evaluation of bistatic algorithms is challenging. A simulation framework has been created to test and evaluate monostatic and bistatic STAP algorithms, mitigating the lack of representative test data. The framework leverages foundational techniques and characteristics to provide a flexible and extensible mechanism for testing and evaluation. Additionally, the design of a new pluggable bistatic Expert System (ES) processor is presented. The ES leverages existing data excision and warping techniques and pairs them with new Range-Based Compensation (RBC) and Clutter Scoring methods to optimize covariance estimation. The simulation framework is used to evaluate the effectiveness of the ES compared to a variety of previously established bistatic processing techniques. The results validate the approach taken in the ES and provide a path for future exploration.

Moving target detection with a passive radar system relies on many competing and coupled variables. When simulating a passive bistatic radar (PBR) system for ground moving target indication (GMTI) a three-dimensional model is critical. The signal path geometry induced from separating the radar receiver and transmitter causes several performance effects that change with location. Since a performance prediction is only as good as the model, the choice of how to model clutter becomes important. Measured data of bistatic clutter shows that the received clutter power depends on scattering angles. Therefore, a new in-plane out-of-plane (IPOP) interpolation model was developed. The IPOP model causes high clutter returns to reside in regions near an in-plane orientation (forward or backward scattering). The model produces a more localized clutter spectrum in angle-Doppler space when compared to monostatic radar. Generally, the stationary transmitter is modeled as a communication emitter due to the availability. These continuous waveforms must be partitioned as pulses spaced at constant intervals over the coherent processing interval (CPI). This diverse pulse train is non-ideal for pulse-Doppler radars. The waveform produces high range sidelobes and causes colored noise to spread in Doppler. It is shown for the first time that these waveform effects can be modeled through a covariance matrix taper (CMT). Choosing an optimal emitter becomes an interesting problem when multiple emitters are present. A common metric for GMTI when using space-time adaptive processing (STAP) is signal-to-interference-plus-noise ratio (SINR). However, SINR changes based off relative geometries, and GMTI depends on where a target's location and two-dimensional velocity maps into angle-Doppler space. Therefore, average SINR, weighted average SINR, minimum SINR, and usable velocity space fraction (UVSF) are the newly developed metrics proposed for down-selecting to an optimal emitter. The choice of metric is extremely dependent on the scenario. Finally, in STAP large clutter discretes (LCDs) can cause either false alarms or missed detections. Ultimately, they contaminate the data, and it is very desirable, yet very hard, to remove LCDs. However, the clutter structure in angle-Doppler space for PBR can offer a benefit for removing an LCD. Due to the fact that bistatic clutter can be more localized in angle-Doppler, the detection and estimation of an LCD can be accomplished for an out-of-plane geometry. Then the LCD can be successfully removed from the data, and new application of spectral estimation techniques have been developed for this purpose.

This thesis investigates the plausibility of implementing a multiple-input single-output (MISO) synthetic aperture radar (SAR) system for space-time adaptive processing (STAP) with a limited data rate requirement of a single receiver. A MISO-SAR system could provide processing flexibility to radar systems such as the Gotcha radar system developed at the Air Force Research Laboratory. Gotcha is an airborne wide-beam multi-mode radar system used to cover a large area for surveillance. In order to apply multiple algorithms to a large amount of data in real time, the data is downlinked to a supercomputer on the ground. STAP is an adaptive filtering technique, which can be used for improved detection of slow moving targets in the presence of clutter. However, STAP is typically implemented using an array of receiving elements, which significantly increases the data rate for downlinking to the ground. While MISO systems are common in communications applications, it is not a common radar system design approach. The MISO system requires additional waveform design considerations in order to obtain orthogonal transmit waveforms. However, the MISO system provides the additional degrees of freedom needed to apply STAP while maintaining a single receiver data rate.