A broad review of publications relevant to nonlinear radar is conducted. The principle-of-operation of nonlinear radar is summarized and applications for this technology are listed. Targets addressed by this type of radar follow a power-series model, and from this model a nonlinear radar range equation is derived. An extensive survey of publicly-available literature, including specifications for systems already tested, guides the design of harmonic radar for finding electronics. The authors have combined a stepped-frequency architecture with harmonic radar to create a system which is capable of imaging and tracking nonlinear targets with very high clutter rejection.
To satisfy the increase in demand for radio frequency (RF) spectrum, recent Federal Communications Commission (FCC) policies permit spectrum sharing between radar and Long-Term Evolution (LTE) communication systems. New cognitive radar systems are promoting spectrum sharing capabilities to mitigate the risk of mutual interference. To develop and test these systems, a realistic communications RF interference (RFI) environment is necessary. This paper describes a system, currently under development, to generate continuous dynamic 4G/LTE RFI for use by radar system designers. The system employs a Vector Signal Transceiver to emulate RFI with various frequency, time-complexity, and power parameters. Many 4G/LTE frames are pre-generated, then, according to the specified parameters, the random frame sequences are generated in real-time. This produces continuous, dynamic, and realistic LTE emissions for a controlled test environment. This work presents implementation details of the LTE emulation system.
This work evaluates the performance of a cognitive radar system which predicts and avoids radio frequency interference (RFI) through an alternating renewal process (ARP) model-based and Markov Decision Process (MDP) approach. As radio frequency (RF) environments grow more crowded, the need for such a system becomes necessary. The cognitive radar monitors the RF activity to train a model for RFI prediction and avoidance. By modeling activity as an alternating renewal process, the stochastic approach calculates the likelihood of interference from measured RFI statistics. Alternatively, the MDP uses reinforcement learning to determine the optimal sequence of decisions given measured RF activity. Both methods eventually select the widest radar transmit bandwidth to minimize interference. The performance of each approach is evaluated by the number of collisions and missed opportunities. A hardware implemented test-bed deploys both methods on a set of synthetic and real measured RFI spectra in real-time to compare performance with the goal of determining when each process is more beneficial (in terms of performance and complexity).
Next generation cognitive radar/radio systems rely on dynamic spectrum access (DSA) to adaptively and ef- ficiently utilize the radio frequency (RF) spectrum. Such technology must detect, predict, and avoid channels occupied by RF interference. Conventional spectrum sensing methods may fail to determine signal occupancy states during transition periods. Predicting RF activity reduces the probability of interference during such transition periods and improves the overall efficiency of DSA schemes. This work employs a one-step ahead prediction approach to determine future busy or idle states through linear support vector regression (SVR). Supervised learning forecasts future signal energy which then acts as a decision statistic to determine occupancy in a sub-band of interest. The scheme’s prediction accuracy is evaluated with respect to input signal-to-noise ratio (SNR) and RF activity as a function of mean busy/idle time. Generalizing RF activity as an alternating renewal process allows exponential random variables to generate simulated data for SVR training and testing. The results show that this approach predicts RF activity with high accuracy over various signal traffic statistics and SNRs. Prediction accuracy is also evaluated with respect to the expected busy/idle transitions given activity statistics.
Nonlinear radar has proven to be a viable means of detecting devices that contain electrical nonlinearities. Electrical nonlinearities are present in dissimilar metals, metal to oxide junctions, semiconductors and more. This paper presents a linear and nonlinear synthetic aperture radar (SAR) system capable of imaging linear and nonlinear targets. The system creates images using data collected from a fixed 16 channel receiver with a single transmitter. A custom 16:1 switching network was developed to collect the SAR data from a 16 antenna receive array. SAR images presented show a nonlinear target placed directly on the ground and imaged in multiple range and cross-range locations. Data is also presented showing the clutter rejection properties of nonlinear radar. Images show that the harmonic radar is able to ignore the strong linear response from a corner reflector, while retaining the nonlinear response from a target.
Dynamic spectrum access (DSA) refers to the adaptive utilization of today’s busy electromagnetic spectrum. Cognitive radio/radar technologies require DSA to intelligently transmit and receive information in changing environments. Predicting radio frequency (RF) activity reduces sensing time and energy consumption for identifying usable spectrum. Typical spectrum prediction methods involve modeling spectral statistics with Hidden Markov Models (HMM) or various neural network structures. HMMs describe the time-varying state probabilities of Markov processes as a dynamic Bayesian network. Neural Networks model biological brain neuron connections to perform a wide range of complex and often non-linear computations. This work compares HMM, Multilayer Perceptron (MLP), and Recurrent Neural Network (RNN) algorithms and their ability to perform RF channel state prediction. Monte Carlo simulations on both measured and simulated spectrum data evaluate the performance of these algorithms. Generalizing spectrum occupancy as an alternating renewal process allows Poisson random variables to generate simulated data while energy detection determines the occupancy state of measured RF spectrum data for testing. The results suggest that neural networks achieve better prediction accuracy and prove more adaptable to changing spectral statistics than HMMs given sufficient training data.
KEYWORDS: Radar, Signal to noise ratio, Transmitters, Cognitive modeling, Detection and tracking algorithms, Data modeling, Databases, Interference (communication), Receivers, Field programmable gate arrays, Data processing, Signal processing, Electromagnetic coupling, Environmental sensing, Commercial off the shelf technology
Clutter and radio frequency interference (RFI) are prevalent issues in the field of radar and are specifically of interest to of cognitive radar. Here, methods for applying and testing the utility of cognitive radar for clutter and RFI mitigation are explored. Using the adaptable transmit capability, environmental database, and general “awareness” of a cognitive radar system (i.e. spectrum sensing, geographical location, etc.), a matched waveform is synthesized that improves the signal-to-clutter ratio (SCR), assuming at least an estimate of the target response and the environmental clutter response are known a prior i. RFI may also be mitigated by sensing the RF spectrum and adapting the transmit center frequency and bandwidth using methods that optimize bandwidth and signal-to-interference plus noise ratio (SINR) (i.e. the spectrum sensing, multi-objective (SS-MO) algorithm). The improvement is shown by a decrease in the noise floor. The above methods’ effectiveness are examined via a test-bed developed around a software defined radio (SDR). Testing and the general use of commercial off the shelf (COTS) devices are desirable for their cost effectiveness, general ease of use, as well as technical and community support, but these devices provide design challenges in order to be effective. The universal software radio peripheral (USRP) X310 SDR is a relatively cheap and portable device that has all the system components of a basic cognitive radar. Design challenges of the SDR include phase coherency between channels, bandwidth limitations, dynamic range, and speed of computation and data communication / recording.
For nonlinear radar, the transmit power required to measure a detectable response from a target is relatively high, and generating that high power is achieved at the cost of linearity. This paper applies the distortion mitigation technique Linearization by Time-Multiplexed Spectrum (LITMUS) to intermodulation radar, a type of nonlinear radar which receives spectral content produced by the mixing of multiple frequencies at a nonlinear target. By implementing LITMUS, an experimental detection system for an intermodulation radar achieves a signal-to-noise ratio up to 20 dB for a total transmit power of approximately 80 mW and nonlinear targets placed at a standoff distance of 2 meters.
The radar range equation for detecting targets using linear radar has been defined and derived many times for many different applications. The nonlinear radar range equation has been defined in the literature but a step by step derivation is lacking and no experimental validation has been shown. This paper starts with a nonlinear system model and provides simulated and experimental validation for the model. Once the model is validated, the model is used to derive the nonlinear radar range equation for nonlinear radar. Key differences between the linear and nonlinear radar range equation will be emphasized.
In many applications of radar systems, detection of targets in environments with heavy clutter and interference can be difficult. It is desired that a radar system should detect targets at a further range as well as be able to detect these targets with very few false positive or negative readings. In a cognitive radar system, there are ways that these negative effects can be mitigated and target detection can be significantly improved. An important metric to focus on for increasing target detectability is the signal-to-clutter ratio (SCR). Cognitive radar offers solutions to issues such as this with the use of a priori knowledge of targets and environments as well as real time adaptations. A feature of cognitive radar that is of interest is the ability to adapt and optimize transmitted waveforms to a given situation. A database is used to hold a priori and dynamic knowledge of the operational environment and targets to be detected, such as clutter characteristics and target radar cross-section (RCS) estimations. Assuming this knowledge is available or can be estimated in real-time, the transmitted waveform can be tailored using methods such as transmission of a spectrum corresponding to the target-to-clutter ratio (TCR). These methods provide significant improvement in distinguishing targets from clutter or interference.
The phase responses of nonlinear-radar targets illuminated by stepped frequencies are studied. Data is presented for an experimental radar and two commercial electronic targets at short standoff ranges. The amplitudes and phases of harmonics generated by each target at each frequency are captured over a 100-MHz-wide transmit band. As in the authors’ prior work, target detection is demonstrated by receiving at least one harmonic of at least one transmit frequency. In the present work, experiments confirm that the phase of a harmonic reflected from a radio-frequency electronic target at a standoff distance is linear versus frequency. Similar to traditional wideband radar, the change of the reflected phase with respect to frequency indicates the range to the nonlinear target.
Today’s military radars are being challenged to satisfy multiple mission requirements and operate in complex, dynamic electromagnetic (EM) environments. They are simultaneously constrained by practical considerations like cost, size, weight and power (SWaP), and lifecycle requirements. Tomorrow’s radars need to be resilient to changing operating environments and capable of doing more with fewer resources. Radar research supports this shift toward more agile and efficient radar systems, and current trends include modular hardware and software development for multi-purpose, scalable radio frequency (RF) solutions. Software-defined radios (SDRs) and other commercial-off-the-shelf (COTS) technology are being used for flexible waveform generation, signal processing, and nontraditional radar applications. Adaptive RF technology, including apertures and other front-end components, are being developed for multi-purpose functionality and resiliency. Together, these research trends will result in a technology framework for more robust future systems that are capable of implementing cognitive processing techniques and adapting their behavior to meet the demands of a congested and contested EM environment.
Last year, we presented the theory behind “instantaneous stepped-frequency, non-linear radar”. We demonstrated through simulation that certain devices (when interrogated by a multi-tone transmit signal) could be expected to produce a multi-tone output signal near harmonics of the transmitted tones. This hypothesized non-linear (multitone) response was then shown to be suitable for pulse compression via standard stepped-frequency processing techniques. At that time, however, we did not have measured data to support the theoretical and simulated results. We now present laboratory measurements confirming our initial hypotheses. We begin with a brief description of the experimental system, and then describe the data collection exercise. Finally, we present measured data demonstrating the accurate ranging of a non-linear target.
The Spectral Analysis Solution (SAS), under development, is a multichannel superheterodyne signal analyzer with the intended applications of radio frequency (RF) research, radar verification, and general purpose spectrum sensing, primarily in the ultra-wideband (UWB) range from ultra high frequency (UHF) to the S-band. The SAS features a wideband channel operating from 100 kHz to 1.8 GHz and eight narrowband channels having adjustable instantaneous bandwidths ranging from 1 MHz to 100 MHz. The wideband channel provides a large picture of the RF spectrum while the narrowband channels allow for high resolution, low noise floor, and high spurious free dynamic range (SFDR) capabilities. An adaptive graphic user interface (GUI) has been implemented for the system that actively pulls and processes the system data in real time. This paper outlines the motivation and theory behind the system along with system validation and implementation results.
Researchers have recently developed radar systems capable of exploiting non-linear target responses to precisely locate targets in range. These systems typically achieve the bandwidth necessary for range resolution through transmission of either a stepped-frequency or chirped waveform. The second harmonic of the reflected waveform is then analyzed to isolate the non-linear target response. In other experiments, researchers have identified certain targets through the inter-modulation products they produce in response to a multi-tone stimulus. These experiments, however, do not exploit the phase information available in the inter-modulation products. We present a method for exploiting both the magnitude and phase information available in the inter-modulation products to create an “instantaneous” stepped frequency, non-linear target response. The new approach enables us to both maintain the unambiguous range dictated by the fundamental, multi-tone separation and obtain the entire target signature from a single transmitted waveform.
In a harmonic radar system design, one of the most important components is the filter used to remove the self-generated harmonics by the high-power transmitter power amplifier, which is usually driven close to its 1-dB compression point. The obvious choice for this filter is a low-pass filter. The low-pass filter will be required to attenuate stop band frequencies with 100 dB attenuation or more. Due to the high degree of attenuation required, multiple low-pass filter will likely be required. Most commercially available low-pass filters are reflective devices, which operate by reflecting the unwanted high frequencies. Cascading these reflective filter causes issues in attenuating stop band frequencies. We show that frequency diplexers are more attractive in place of reflective low-pass filters as they are able to terminate the stop band frequencies as opposed to reflecting them.
Radio-frequency (RF) electronic targets, such as man-portable electronics, cannot be detected by traditional linear radar because the radar cross section of those targets is much smaller than that of nearby clutter. One technology that is capable of separating RF electronic targets from naturally-occurring clutter is nonlinear radar. Presented in this paper is the evolution of nonlinear radar at the United States Army Research Laboratory (ARL) and recent results of short-range over-the-air harmonic radar tests there. For the present implementation of ARL’s nonlinear radar, the transmit waveform is a chirp which sweeps one frequency at constant amplitude over an ultra-wide bandwidth (UWB). The receiver captures a single harmonic of this entire chirp. From the UWB received harmonic, a nonlinear frequency response of the radar environment is constructed. An inverse Fourier Transform of this nonlinear frequency response reveals the range to the nonlinear target within the environment. The chirped harmonic radar concept is validated experimentally using a wideband horn antenna and commercial off-the-shelf electronic targets.
This paper presents synthetic aperture radar (SAR) images of linear and nonlinear targets. Data are collected using a linear/nonlinear step frequency radar. We show that it is indeed possible to produce SAR images using a nonlinear radar. Furthermore, it is shown that the nonlinear radar is able to reduce linear clutter by at least 80 dB compared to a linear radar. The nonlinear SAR images also show the system’s ability to detect small electronic devices in the presence of large linear clutter. The system presented here has the ability to completely ignore a 20-inch trihedral corner reﬂector while detecting a RF mixer with a dipole antenna attached.
Microwave power amplifiers often operate in the nonlinear region to maximize efficiency. However, such operation inevitably produces significant harmonics at the output, thereby degrading the performance of the microwave systems. An automated method for canceling harmonics generated by a power amplifier is presented in this paper. Automated tuning is demonstrated over 400 MHz of bandwidth with a minimum cancellation of 110 dB. The intended application for the harmonic cancellation is to create a linear radar transmitter for the remote detection of non-linear targets. The signal emitted from the non-linear targets is often very weak. High transmitter linearization is required to prevent the harmonics generated by the radar itself from masking this weak signal.
In this paper, spectrum sensing techniques are explored for nonlinear radar. These techniques use energy detection to identify an unoccupied receive frequency for nonlinear radar. A frequency is considered unoccupied if it satisfies the following criteria: 1) for a given frequency of interest, its energy must be below a predetermined threshold; 2) the surrounding energy of this frequency must also be below a predetermined threshold. Two energy detection techniques are used to select an unoccupied frequency. The first technique requires the fast Fourier transform and a weighting function to test the energy in neighboring frequency bins; both of these procedures may require a high degree of computational resources. The second technique uses multirate digital signal processing and the fast binary search techniques to lower the overall computational complexity while satisfying the requirements for an unoccupied frequency.
RF electronic targets cannot be detected by traditional linear radar because their radar cross sections are much smaller than that of nearby clutter. One technology that is capable of separating RF electronic targets from clutter, however, is nonlinear radar. Presented in this paper is a combination of stepped-frequency ultra-wideband radar with nonlinear detection. By stepping the transmit frequency across an ultra-wide bandwidth and recording the amplitude and phase of the harmonic return signal, a nonlinear frequency response of the radar environment is constructed. An inverse Fourier transform of this response reveals the range to a nonlinear target.
An increasingly cluttered electromagnetic environment (EME) is a growing problem for radar systems. This problem is
becoming critical as the available frequency spectrum shrinks due to growing wireless communication device usage and
changing regulations. A possible solution to these problems is cognitive radar, where the cognitive radar learns from the
environment and intelligently modifies the transmit waveform. In this paper, a cognitive nonlinear radar processing
framework is introduced where the main components of this framework consist of spectrum sensing processing, target
detection and classification, and decision making. The emphasis of this paper is to introduce a spectrum sensing
processing technique that identifies a transmit-receive frequency pair for nonlinear radar. It will be shown that the
proposed technique successfully identifies a transmit-receive frequency pair for nonlinear radar from data collected from
Nonlinear radar exploits the electronic response from a target whose reflected frequencies are different from those transmitted. Reception of frequencies that are not part of the transmitted probe distinguishes the received signal from a linear return produced by clutter and indicates the presence of electronics. Presented in this paper is a type of nonlinear radar that transmits multiple frequencies and listens for a harmonic of these frequencies as well as other frequencies near that harmonic. A laboratory test-bed has been constructed to demonstrate the multitone radar concept. Measurements of nonlinear responses from RF devices probed by multiple tones are reported.
KEYWORDS: Target detection, Radar, Signal to noise ratio, Spectrum analysis, Detection and tracking algorithms, Telecommunications, Signal generators, Algorithm development, Signal detection, Environmental sensing
Providing situational awareness to the warfighter requires radar, communications, and other electronic systems that
operate in increasingly cluttered and dynamic electromagnetic environments. There is a growing need for cognitive RF
systems that are capable of monitoring, adapting to, and learning from their environments in order to maintain their
effectiveness and functionality. Additionally, radar systems are needed that are capable of adapting to an increased
number of targets of interest. Cognitive nonlinear radar may offer critical solutions to these growing problems. This
work focuses on ongoing efforts at the U.S. Army Research Laboratory (ARL) to develop a cognitive nonlinear radar
test-bed. ARL is working toward developing a test-bed that uses spectrum sensing to monitor the RF environment and
dynamically change the transmit waveforms to achieve detection of nonlinear targets with high confidence. This work
presents the architecture of the test-bed system along with a discussion of its current capabilities and limitations. A brief
outlook is presented for the project along with a discussion of a future cognitive nonlinear radar test-bed.
The Army Research Laboratory (ARL) has, in the past, demonstrated the effectiveness of low frequency,
ultrawideband radar for detection of slow-moving targets located behind walls. While these initial results
were promising, they also indicated that sidelobe artifacts produced by moving target indication (MTI)
processing could pose serious problems. Such artifacts induced false alarms and necessitated the
introduction of a tracker stage to eliminate them. Of course, the tracker algorithm was also imperfect, and
it tended to pass any persistent, nearly collocated false alarms.
In this work we describe the incorporation of a sidelobe-reduction technique-the randomized linear
receiver array (RA)-into our MTI processing chain. To perform this investigation, we leverage data
collected by ARL's synchronous impulse reconstruction (SIRE) radar. We begin by calculating MTI
imagery using both the non-random and randomized array methods. We then compare the sidelobe levels
in each image and quantify the differences. Finally, we apply a local-contrast target detection algorithm
based on constant false alarm rate (CFAR) principles, and we analyze probabilities of detection and false
alarm for each MTI image.
KEYWORDS: Radar, Image processing algorithms and systems, Synthetic aperture radar, Image segmentation, Image processing, Image acquisition, Data processing, Signal processing, Antennas, Global Positioning System
Researchers at the U.S. Army Research Laboratory (ARL) designed and fabricated the Synchronous Impulse
REconstruction (SIRE) radar system in an effort to address fundamental questions about the utilization of low
frequency, ultrawideband (UWB) radar. The SIRE system includes a receive array comprising 16 receive channels,
and it is capable of operating in either a forward-looking or a side-looking mode. When operated in side-looking
mode, it is capable of producing high-resolution Synthetic Aperture Radar (SAR) data. The SAR imaging
algorithms, however, initially operated under the assumption that the vehicle followed a nearly linear trajectory
throughout the data collection. Under this assumption, the introduction of vehicle path nonlinearities distorted the
processed SAR imagery. In an effort to mitigate these effects, we first incorporated segmentation routines to
eliminate highly non-linear portions of the path. We then enhanced the image formation algorithm, enabling it to
process data collected from a non-linear vehicle trajectory.
We describe the incorporated segmentation approaches and compare the imagery created before and after their
incorporation. Next, we describe the modified image formation algorithm and present examples of output imagery
produced by it. Finally, we compare imagery produced by the initial segmentation algorithm to imagery produced by
the modified image-formation algorithm, highlighting the effects of segmentation parameter variation on the final
Moving target indication (MTI) algorithms often operate within a relatively narrow frequency band
relying on Doppler processing to detect moving targets at long standoff ranges. At these standoff ranges,
received wavefronts impinging on a linear array can be considered planar, enabling implementation of a
variety of phase-based beam-forming techniques. At near ranges, however, the plane-wave assumption no
longer holds. We describe enhancements to an impulse-based, low-frequency, ultra-wideband, moving-target
imaging system for near-range, through-the-wall MTI. All MTI image processing is performed in
the time domain using a change detection (CD) paradigm. We discuss how MTI image quality can be
increased through the introduction of randomized linear arrays. After describing the process in detail, we
present results obtained using data collected by an impulse-based, low frequency, ultra-wideband system.
A local text alignment algorithm is introduced in this work for synchronizing transcripts. The proposed algorithm can be used for any transcript alignment process where high computational complexity is a concern. Dynamic programming is typically used to align a set of transcripts: however, the computational complexity of dynamic programming is high. To reduce the computational complexity, a local dynamic programming algorithm is introduced that aligns subsections of the transcripts. Aligning subsections of the transcripts greatly reduces the information needed for accurate synchronization. The information is reduced because it is not necessary to compare all words between the two transcripts. For example, words at the beginning of one transcript would not be compared to the words at the end of the other transcript. The subsection size is dependent on the total number of alignment errors between the transcripts. It is shown that the computational complexity of the proposed local dynamic programming algorithm is greatly reduced while preserving alignment accuracy.
Previously, we developed a moving target indication (MTI) processing approach to detect and track slow-moving targets
inside buildings, which successfully detected moving targets (MTs) from data collected by a low-frequency, ultra-wideband
radar. Our MTI algorithms include change detection, automatic target detection (ATD), clustering, and
tracking. The MTI algorithms can be implemented in a real-time or near-real-time system; however, a person-in-the-loop
is needed to select input parameters for the clustering algorithm. Specifically, the number of clusters to input into the
cluster algorithm is unknown and requires manual selection. A critical need exists to automate all aspects of the MTI
processing formulation. In this paper, we investigate two techniques that automatically determine the number of clusters:
the adaptive knee-point (KP) algorithm and the recursive pixel finding (RPF) algorithm. The KP algorithm is based on a
well-known heuristic approach for determining the number of clusters. The RPF algorithm is analogous to the image
processing, pixel labeling procedure. Both algorithms are used to analyze the false alarm and detection rates of three
operational scenarios of personnel walking inside wood and cinderblock buildings.
Remote detection and characterization of wireless devices in an environment is a topic of growing importance.
Characterization of a wireless device is useful in many applications. An example of this is in the testing of FCC Part 15
devices. These devices must adhere to strict guidelines in regards to RF interference. Compliance can be verified by
using forensic techniques to classify and characterize the returned signal. We present a framework for remote detection
and forensic characterization of RF devices using specially designed probe signals. This framework can be applied to a
broad range of devices and models. Probe signals, device models, feature selection, classifier design are described. For
the device model we introduce a method for simulating a non-linearity in the RF system based on a known diode model.
Experimental results are given to verify our approach.
This paper presents a time-domain, Moving-Target-Indication (MTI) processing formulation for detecting slow-moving
personnel behind walls. The proposed time-domain MTI processing formulation consists of change detection and
tracking algorithms. We demonstrate the effectiveness of the MTI processing formulation using data collected by the
Army Research Laboratory's (ARL's), Ultra-Wideband (UWB), Synchronous Impulse Reconstruction (SIRE) radar.
During the collection of the data, the SIRE radar remains stationary and is positioned broadside to the wall and 38
degrees off the broadside position. We have collected data for multiple operational scenarios including: personnel
walking inside wood and cinderblock structures, personnel walking in linear and non-linear trajectories, and multiple
personnel walking within the building structure. We analyze the characteristics of moving target signatures for the
multiple operational scenarios and describe the detection and tracking algorithms implemented to exploit them.
The Army Research Laboratory (ARL) has recently developed the ground-based synchronous impulse reconstruction
(SIRE) radar - a low-frequency radar capable of exploiting both a real antenna array and along-track integration
techniques to increase the quality of processed imagery. We have already demonstrated the system's utility by imaging
static scenes. In this paper we address the moving target indication (MTI) problem, and we demonstrate the impulse-based
system's ability to both detect and locate slowly moving targets. We begin by briefly describing the SIRE system
itself as well as the system configuration utilized in collecting the MTI data. Next we discuss the signal processing
techniques employed to create the final MTI image. Finally, we present processed imagery illustrating the utility of the
Change detection provides a powerful tool for detecting the introduction of weapons or hazardous materials into an area
under surveillance, as demonstrated in past work carried out at the Army Research Laboratory (ARL). This earlier work
demonstrated the potential for detecting recently emplaced surface landmines using an X-Band, synthetic aperture radar
(SAR) sensor. Recent experiments conducted at ARL have extended these change detection results to imagery collected
by the synthetic impulse reconstruction (SIRE) radar - a lower-frequency system developed at ARL. In this paper we
describe the algorithms adopted for this change detection experiment and present results obtained by applying these
algorithms to the SIRE data set. Results indicate the potential for utilizing systems such as the SIRE as surveillance
Radar systems have long been recognized as an effective tool for detecting moving targets--a problem commonly referred to as moving target indication (MTI). Recent advances, including Space Time Adaptive Processing (STAP), allow for even more precise determination of a target's location relative to the radar. Still, most of these methods approach MTI from the point of view of parameter estimation, and this sort of an approach can become problematic when the target speed is low and its associated Doppler frequency is near zero. In such cases the target signature is masked by the stationary, background clutter. Another potential drawback to STAP techniques arises from the fact that they require a relatively large number of receive channels, adding additional complexity to the radar system hardware.
In this paper we present a moving-target-indication (MTI) technique that is based on a change detection paradigm. That is, rather than estimating the Doppler frequency associated with a target's motion, we propose to detect subtle differences between simultaneously collected, complex SAR images. We use simulated data to illustrate the feasibility of the approach under several different operating scenarios.
The process of identifying speakers in a news program is difficult using only text information. We propose a system that will first perform text and video processing separately to identify the start of speech of a speaker. These start of speech locations are aligned and used to identify a change of speaker in the program. An analysis is performed to identify the contribution of the text and video information. It will be be shown that the change of speaker locations identified by our alignment algorithm is more accurate then either mode individually.
The production of closed captions is an important but expensive process in video broadcasting. We propose a method to generate highly accurate off-line captions efficiently. Our system uses text alignment to synchronize program transcripts obtained for a video program with text produced by an automatic speech recognition (ASR) system. We will also describe the accuracy in both closed-caption text and the ASR output for a number of news programs and provide a detailed analysis of the errors that occur.