Henna Perälä

On efficient characterization of radar targets with scatterer sets for target recognition using commercial ray tracing software

The model-based radar target recognition requires the ability to simulate target signatures with adequate accuracy and computational speed. This paper proposes an entirely novel approach to utilizing readily available ray tracing software in the target characterization. The objective of the proposed approach is to give the target a compact description, from which arbitrary radar response can be simulated with efficiency. Certain essential computational aspects, such as ray density reduction and frequency-independency, are discussed. The simulated radar response is compared with a measurement, covering the case of a single frequency, high range resolution profiles, and inverse synthetic aperture radar images.

Incorporating a stochastic model of the target orientation into a momentary RCS distribution

The radar cross section (RCS) of a target describes its reflectivity towards the radar. It is a deterministic quantity-specific to the aspect angle of the target and the properties of the radar signal-and as such can be predicted with various methods when provided with an adequate physical model of the target. However, stochasticity is introduced when considering a target during flight and several factors increasing the uncertainty-typically about the location and orientation of the target in relation to the observing radar-have to be taken into account. It is possible to model these uncertainties when adequate information about physical properties of the target as well as about its flight path and motion mode is available. In this paper, we propose a method for incorporating such knowledge about the flight dynamics into the estimation of a momentary distribution for the RCS. We use a simulated trajectory to demonstrate the method and present comparisons to the Swerling I model.

Performance evaluation of radar NCTR using the target aspect and signature

In this paper, we present a framework for assessing and comparing the performance of non-cooperative target recognition methods that are based on radar signatures. The proposed framework aspires to quantify the performance considering the dynamic nature of the operating conditions. When evaluating the recognition performance, the training data is generated according to the operational scenario which specifies for instance the values for the true aspect angle, at which the target is observed, and the ability to estimate it, e.g. by a radar tracker. The performance of different classifiers can be assessed and compared considering the varying ability to observe the location and the orientation of the target. The proposed framework is demonstrated with the radar cross section as the feature for classification using values on a single frequency and high range resolution profiles.

Converting a 3D surface into a set of directional scatterers for high-resolution radar response simulation

It is practical and efficient to simplify targets to point scatterers in radar simulations. With low-resolution radars, the radar cross section (RCS) is a sufficient feature to characterize the scattering properties of a target. However, the RCS totals the target scattering properties to a scalar value for each aspect angle. Thus, a more detailed representation of the target is required with high-resolution radar techniques, such as Inverse Synthetic-Aperture Radar (ISAR). In straightforward simulation scenarios, high-resolution targets have been modeled placing identical point scatterers in the shape of the target, or with a few dominant point scatterers. As extremely simple arrangements, these do not take the self-shadowing into account and are not realistic enough for high demands. Our radar response simulation studies required a target characterization akin to RCS, which would also function in highresolution cases and take the self-shadowing and multiple reflections into account. Thus, we propose an approach to converting a 3-dimensional (3D) surface into a set of scatterers with locations, orientations, and directional scattering properties. The method is intended for far field operation, but could be adjusted for use in the near field. It is based on ray tracing which provides the self-shadowing and reflections naturally. In this paper, we present ISAR simulation results employing the proposed method. The constructed scatterer set is scalable for different wavelengths enabling the fast production of realistic simulations including authentic RCS scattering center formation. This paper contributes to enhancing the reality of the simulations, yet keeping them manageable and computationally reasonable.

A RCS model of complex targets for radar performance prediction

The objective of the radar performance prediction is to compute quantities of interest concerning the ability of the radar to observe its surroundings. Besides the properties of the radar system, the performance is affected by the target, whose radar cross section (RCS) is one of the predominant factors. The performance prediction is usually performed in relation to the target RCS characterized by a constant value or a particular statistical distribution. Such representations generalize real-life complex targets rendering them unsuitable for some objectives since the RCS is significantly influenced by the target aspect angle and is inherently stochastic by nature. Thus, a more dynamic description may be valuable e.g. for analyzing the radar performance on a flight path of interest. We propose representing the RCS with a histogram that includes such dynamic properties and is suitable for considering the target in different ways for performance prediction: in a more general manner or dependent on its aspect angle. We consider the case of traditional RCS with low spatial resolution and demonstrate the proposed approach through the probability of detection computed for a generic surveillance radar.

ESPRESS—On efficient bistatic characterization of radar targets

In the modern radar target recognition, the model-based approach offers a flexible and computationally efficient way to characterize targets, since establishing an adequate target signature collection especially with bistatic measurements is impractical. Simulating such an extensive collection is arduous as well. This paper proposes a new method for the bistatic characterization of radar targets and radar response simulation: ESPRESS (Electromagnetic Signature Production from Renders Exploiting Scatterer Sets). We have implemented it entirely with commercial off-the-shelf (COTS) software. Our objective is to give the target a compact description, from which radar response-with arbitrary radar frequency and bandwidth, as well as transmitter and receiver positions-can be simulated efficiently. In this paper, we demonstrate that ESPRESS has the computational speed and adequate accuracy required in model-based radar target recognition.

An approach to enhanced fidelity of airborne radar site-specific simulation

Geographical information systems (GIS) have been the base for radar ground echo simulations for many years. Along with digital elevation model (DEM), present GIS contain characteristics of terrain. This paper proposes a computationally sensible simulation procedure to produce realistic radar terrain signatures in a form of raw data of airborne pulse Doppler radar. For backscattering simulation, the model of the ground is based on DEM and built with point-form backscattering objects. In addition to the usual DEM utilization for xyz coordinates and shadowed region calculation, we assume that each data point in GIS describes several scatterers in reality. Approaching the ground truth, we distribute individual scatterers with adjustable attributes to produce authentic response of areas such as sea, fields, forests, and built-up areas. This paper illustrates the approach through an airborne side-looking synthetic aperture radar (SAR) simulation. The results prove the enhanced fidelity with realistic SAR image features.

Classification of radar data by detecting and identifying spatial and temporal anomalies

For some time, applying the theory of pattern recognition and classification to radar signal processing has been a topic of interest in the field of remote sensing. Efficient operation and target indication is often hindered by the signal background, which can have similar properties with the interesting signal. Because noise and clutter may constitute most part of the response of surveillance radar, aircraft and other interesting targets can be seen as anomalies in the data. We propose an algorithm for detecting these anomalies on a heterogeneous clutter background in each range-Doppler cell, the basic unit in the radar data defined by the resolution in range, angle and Doppler. The analysis is based on the time history of the response in a cell and its correlation to the spatial surroundings. If the newest time window of response in a resolution cell differs statistically from the time history of the cell, the cell is determined anomalous. Normal cells are classified as noise or different type of clutter based on their strength on each Doppler band. Anomalous cells are analyzed using a longer time window, which emulates a longer coherent illumination. Based on the decorrelation behavior of the response in the long time window, the anomalous cells are classified as clutter, an airplane or a helicopter. The algorithm is tested with both experimental and simulated radar data. The experimental radar data has been recorded in a forested landscape.

Merging Radar Data with Geographic Data for Visual Land Clutter Source Recognition

In the recent years, radar land clutter modelling and processing have been aided with Geographic Information Systems (GIS) and geodata in a few recognised researches such as in the Lincoln Laboratory. In our clutter research, one aspect is to study the possibilities of using GIS in clutter classification in Finnish environment. Since the automation of this process causes inaccurate results and a need to identify and label various types of land clutter sources through geographic data (geodata) exists, we propose an approach based on the visual interpretation of clutter. We have created a graphical visualisation tool for merging geodata with radar data interactively, including an option to select the shown type(s) of geodata. The source identification is based on the visual observation of the output. The tool can also be utilised when verifying simulated data. In an example case, we have used the following geodata items: a base map, a terrain model, a database of tall structures, and a digital elevation model, but other types of geodata can be used as well. Although the potential to enhance the model is higher when more types of geodata are utilised, even with few carefully selected geodata items, clutter sources can be recognised adequately. This paper presents an illustrative demonstration using an air surveillance radar recording. This visual approach with the data merging tool has been useful, and the results have verified the practicability. The contribution of this paper focuses on supporting clutter classification research and improving the understanding of land clutter.

Estimating dynamics of heavily fluctuating radar responses: a land clutter application and experimental results

The strength of radar response varies considerably. In this regard, the dynamic range of most receivers is not sufficient enough to operate optimally. Due to this fact, radar signal may represent only a fraction of the real backscattering phenomena. One way to solve the problem is to use automatic gain control (AGC). It helps to prevent the saturation of responses but inflicts performance degradation on subsequent radar signal processing. The same problem with dynamic range exists in other fields of sensing as well. For example, a solution in digital photography is to use various exposure times to determine the most appropriate one for the current conditions. In this paper, a corresponding approach is proposed for analyzing radar responses. The method requires measurements of a selected area to be performed with various gains, and the resulting dynamic ranges should overlap partially. The use of a linear receiver ensures that both the power and the coherent phase statistics can be extracted from the data. Using the proposed approach, a few distributions derived from extensive land clutter recordings from Finnish landscape are presented.