James Robertson

Artificial neural networks for acoustic target recognition

Acoustic sensors can be used to detect, track and identify non-line-of-sight targets passively. Attempts to alter acoustic emissions often result in an undesirable performance degradation. This research project investigates the use of neural networks for differentiating between features extracted from the acoustic signatures of sources. Acoustic data were filtered and digitized using a commercially available analog-digital convertor. The digital data was transformed to the frequency domain for additional processing using the FFT. Narrowband peak detection algorithms were incorporated to select peaks above a user defined SNR. These peaks were then used to generate a set of robust features which relate specifically to target components in varying background conditions. The features were then used as input into a backpropagation neural network. A K-means unsupervised clustering algorithm was used to determine the natural clustering of the observations. Comparisons between a feature set consisting of the normalized amplitudes of the first 250 frequency bins of the power spectrum and a set of 11 harmonically related features were made. Initial results indicate that even though some different target types had a tendency to group in the same clusters, the neural network was able to differentiate the targets. Successful identification of acoustic sources under varying operational conditions with high confidence levels was achieved.

The Impact of the Data Archiving File Format on Scientific Computing and Performance of Image Processing Algorithms in MATLAB Using Large HDF5 and XML Multimodal and Hyperspectral Data Sets

Scientists require the ability to effortlessly share and process data collected and stored on a variety of computer platforms in specialized data storage formats. Experiments often generate large amounts of raw and corrected data and metadata, which describes and characterizes the raw data. Scientific teams and groups develop many formats and tools for internal use for specialized users with particular references and backgrounds. Researchers need a solution for querying, accessing, and analyzing large data sets of heterogeneous data, and demand high interoperability between data and various applications (Shasharina et al., 2007; Shishedjiev et al., 2010). Debate continues regarding which data format provides the greatest transparency and produces the most reliable data exchange. Currently, Extensible Markup Language (XML) and Hierarchical Data Format 5 (HDF5) formats are two solutions for sharing data. XML is a simple, platform-independent, flexible markup meta-language that provides a format for storing structured data, and is a primary format for data exchange across the Internet (McGrath, 2003). XML data files use Document Type Definitions (DTDs) and XML Schemas to define the data structures and definitions, including data formatting, attributes, and descriptive information about the data. A number of applications exist that use XML-based storage implementations for applications, including radiation and spectral measurements, simulation data of magnetic fields in human tissues, and describing and accessing fusion and plasma physics simulations (Shasharina et al., 2007; Shishedjiev et al., 2010). HDF5 is a data model, library, and file format for storing and managing data. HDF5 is portable and extensible, allowing applications to evolve in their use of HDF5 (HDF Group). HDF5 files provide the capability for self-documenting storage of scientific data in that the HDF5 data model provides structures that allow the file format to contain data about the file structure and descriptive information about the data contained in the file (Barkstrom, 2001). Similar to XML, numerous applications using the HDF5 storage format exist, such as fusion

Cloud-based security architecture supporting Army Research Laboratory's collaborative research environments

Designing, operating and maintaining any system in today’s complex Information Technology (IT) environments requires understanding and mitigating various levels of risk to the organization or to individuals associated with the operation of a system. How this risk is managed is critical to the organizations success. Multiple assets, technologies and skilled personnel are needed to select an appropriate set of security controls needed to protect individuals, operations and assets of the organization, and build or refine a security architecture based on best practices and risk management frameworks to adapt to ever changing threats. Developing secure applications and collaborating within a cloud-based environment adds both challenges and mitigations. Cloud environments are scalable and with the proper design and use can be secure and cost-effective. Special attention, processes and techniques associated with authentication and authorization require the formation and maintenance of users, groups, roles and policies. The U.S. Army Research Laboratory (ARL) provides Americas Soldiers the technological edge through scientific research, technology development, and analysis. ARL provides scientific and technological innovation in a variety of technical disciplines, through direct in-house laboratory efforts and joint programs with government, industry, and academia. ARL’s Open Campus is a collaborative endeavor, with the goal of building a science and technology ecosystem that will encourage groundbreaking advances in basic and applied research areas of relevance to the Army. Through the Open Campus framework, ARL scientists and engineers work collaboratively and side-by-side with visiting scientists in ARLs facilities, and as visiting researchers at collaborators institutions. Through the relationships formed and the availability of secure, dynamic and scalable environments, rapid development, sharing and transition of technologies is possible. This technical paper proposes a cloud-based security architecture to support multiple Open Campus Initiatives at the Army Research Laboratory including the Sensor Information Testbed COllaborative Research Environment (SITCORE) and the Automated Online Data Repository (AODR). These initiatives create a highly-collaborative research laboratory and testbed environment focused on sensor data and information fusion. Coupling the existing Open Campus Initiatives with an additional cloud-based architecture allowing encrypted communication, authentication, and on-demand access provides a scalable and secure environment supporting the data, algorithm, and collaborative needs of scientists, researchers and entrepreneurs.

Signal and image processing algorithm performance in a virtual and elastic computing environment

The U.S. Army Research Laboratory (ARL) supports the development of classification, detection, tracking, and localization algorithms using multiple sensing modalities including acoustic, seismic, E-field, magnetic field, PIR, and visual and IR imaging. Multimodal sensors collect large amounts of data in support of algorithm development. The resulting large amount of data, and their associated high-performance computing needs, increases and challenges existing computing infrastructures. Purchasing computer power as a commodity using a Cloud service offers low-cost, pay-as-you-go pricing models, scalability, and elasticity that may provide solutions to develop and optimize algorithms without having to procure additional hardware and resources. This paper provides a detailed look at using a commercial cloud service provider, such as Amazon Web Services (AWS), to develop and deploy simple signal and image processing algorithms in a cloud and run the algorithms on a large set of data archived in the ARL Multimodal Signatures Database (MMSDB). Analytical results will provide performance comparisons with existing infrastructure. A discussion on using cloud computing with government data will discuss best security practices that exist within cloud services, such as AWS.

The U.S. Army Research Laboratory (ARL) multimodal signature database (MMSDB) advanced data storage solutions and security of data over the web

The U.S. Army Research Laboratory (ARL) archives vast amounts of data requiring a secure, portable file format, along with a versatile software library for storing and accessing its data. Hierarchical Data Format 5 (HDF5) is a popular, general-purpose library and open-source file format designed for archiving data, and providing extreme interoperability and data encryption for secure accessibility. This paper will provide an overview of the current state of effectively integrating encryption algorithms into HDF5 datasets, along with possible applications, expectations, and limitations, including a discussion on creating a framework for dissemination of sensitive data over the Web.

Advances in the design, development, and deployment of the U.S. Army Research Laboratory (ARL) multimodal signatures database

Recent advances in the design, development, and deployment of U.S. Army Research Laboratorys (ARL) Multimodal Signature Database (MMSDB) create a state-of-the-art database system with Web-based access through a Web interface designed specifically for research and development. Tens of thousands of signatures are currently available for researchers to support their algorithm development and refinement for sensors and other security systems. Each dataset is stored in (Hierarchical Data Format 5 (HDF5) format for easy modeling and storing of signatures and archived sensor data, ground truth, calibration information, algorithms, and other documentation. Archived HDF5 formatted data provides the basis for computational interoperability across a variety of tools including MATLAB, Octave, and Python. The database has a Web-based front-end with public and restricted access interfaces, along with 24/7 availability and support. This paper describes the overall design of the system, and the recent enhancements and future vision, including the ability for researchers to share algorithms, data, and documentation in the cloud, and providing an ability to run algorithms and software for testing and evaluation purposes remotely across multiple domains and computational tools. The paper will also describe in detail the HDF5 format for several multimodal sensor types.

The impact of the data archiving file format on the sharing of scientific data for use in popular computational environments

The U.S. Army Research Laboratory (ARL) conducted an initial study on the performance of XML and HDF5 in three popular computational software environments, MATLAB, Octave, and Python, all of which use high-level scripting languages and computational software tools designed for computational processing. Although usable for sharing and exchanging data, the initial results of the study indicated XML has clear limitations in a computational environment. Popular computational tools are unable to handle very large XML formatted files, thus limiting processing of large XML archived data files. We show the breakdown points of XML formatted files for various popular computational tools and explore the performance dependencies of XML and HDF5 formatted files in popular computational environments on the hardware, operating system, and mathematical function. This study also explores the inverse file size relationship between HDF5 and XML data files. Several organizations, including ARL, use both XML and HDF5 for archiving and exchanging data. XML is best suited for storing light data (such as metadata) and HDF5 is best suited for storing heavy scientific data. Integrating and using both XML and HDF5 for data archiving offers the best solution for data providers and consumers to share information for computational and scientific purposes.

Multimodal signature file formats and performance in computational environments

The dissemination formats for multimodal signature data generally favor either formats with extreme interoperability, such as Extensible Markup Language (XML), portable binary formats with very high levels of interoperability, such as Hierarchical Data Format 5 (HDF5), or proprietary binary formats, often optimized for a specific application but which offer very low levels of interoperability across various computer platforms. The lack of Free and Open Source Software (FOSS) tools for proprietary or application specific file formats, tends to make such formats inappropriate for sharing signature data across organizations such as U.S. Army Research Laboratory (ARL), Signatures Support Program (SSP), and other government agencies. Sharing signature data for computational purposes is of extreme interest to the scientific community. An initial study of similar signatures in two different popular data file formats (HDF5 and XML) and using three popular computational environments (MATLAB, Octave, and Python) reveals definite advantages of HDF5 over XML, especially for larger data sets. HDF5 provides several key benefits for scientific applications without sacrificing interoperability across many computer platforms. The combination of HDF5 and XML for dissemination of signature data and information may yield the best solution for data consumers and providers.

Detection of degraded target signatures: statistical versus neural networks

Pattern recognition applications require algorithms be optimized to provide accurate and reproducible target identification. Approaches usually incorporate a combination preprocessing, feature extraction, and classification algorithms whose parameters have been adjusted for the best performance against a particular set of images. With the variety of neural network and statistical techniques available at each of these processing steps, choosing the correct algorithms for a particular application may be difficult. A Pattern Recognition Workstation (PRW) has been developed to assist in the selection of these algorithms. The workstation provides a variety of image degradation techniques to assist the user in assessing the performance of algorithms as a function of obscuration, noise levels, scale and rotation. Initial results are reported from preprocessors including the Contrast-Orientation-Ratio- Threshold-Maximum (CORT-X), Sobel and Laplacian, feature extractors including the Gabor Transform, Invariant Moments, and Fourier-Log-Polar Transform, and classifiers including Backpropagation and Bayes decision theory. The resulting class decision statistics are presented to assess robustness with respect to obscuration and noise levels.

Prototype neural network pattern recognition testbed

Recent successes ofneural networks has led to an optimistic outlook for neural network applications to image processing(IP). This paperpresents a general architecture for performing comparative studies of neural processing and more conventional IF techniques as well as hybrid pattern recognition (PR) systems. Two hybrid PR systems have been simulated each of which incorporate both conventional and neural processing techniques.