Ben A. Abbott

Precise positioning with wireless sensor nodes: Monitoring natural hazards in all terrains

Prediction, assessment, and mitigation of surface-affecting natural hazard processes such as landslides, avalanches, earthquakes, and floods call upon geoscientists to rapidly deploy instruments and accurately characterize these earth processes, often with little lead time and under dangerous working conditions. Affected areas may have heavy tree canopies, or high atmospheric dust loads (volcanic eruptions), precluding the use of traditional location techniques like Global Positioning System (GPS). The proliferation of inexpensive radio systems provides a technology that has the potential to redefine the approach to rapid characterization of hazardous earth processes. The research effort described in this paper developed and demonstrated an inexpensive, cooperative radar-like technology for precise distance measurement between intelligent radio nodes.

Model-Based Cyber Security

This paper presents an approach for automatically verifying the correctness of cyber security applications through formal analysis guided by hierarchical models of the network, its applications, and potential attacks. This work is motivated by the need for a more intuitive, automated systems-level approach to determining the overall security characteristics of a large network. Given the complex nature of security tools and their general lack of interoperability, it is difficult for system designers to make definitive statements about the nature of their network defense. Our work focuses on creating an environment in which security experts can model the security aspects of complex networks using a graphical notation that is intuitive and natural for them, then automatically perform security activities such as formally verifying the safety of the network against known threats and exploring the network design for potential vulnerabilities. The environment is designed to utilize third party tools for performing these activities and concentrates on integration of these tools within a common modeling framework

Mapping borehole-accessed karst solutional features and culvert conduits using remote sensor technology

Prior papers have described prototype sensors that were developed to autonomously map pathway, flow velocity, and dimensions as they flow through karst conduits. The prototype sensors are equipped with sonar and magnetometers to measure conduit morphology and orientation. The sensors are developed to be approximately neutrally buoyant but have been equipped with a propulsion system to enable the sensors to negotiate around impediments in the flow channel and avoid stalling at the walls of the conduit or cave. Data collected during an excursion are downloaded from the sensor upon completion of the survey mission. An autonomous sensor was successfully used to characterize a segment in Honey Creek Cave, a wet cave in south-central Texas. Sonar proved to be effective in measuring the cave dimensions and the velocity of flow. A magnetometer was used to orient the pathway taken by the sensor. Together, these data provided a representative reproduction of the oriented morphology of the wet cave. Two variations of the initial generation of sensors have been developed to meet the requirements of projects funded by the United States Army Corps of Engineers for mapping borehole-accessed karst solution cavities and by the Federal Highway Administration for mapping, monitoring, and diagnosing damage to roadway culverts. The first variation is tethered to map karst voids intersected by a drill hole but where discharge to a spring is not anticipated. The second features an enhanced sonar scheme to overcome the extreme multipath environments found inside a partially filled metal culvert and to provide localization information in a magnetically shielded environment.

Context-based object recognition

Several methods have been developed for context-based object recognition within aerial imagery. These methods were inspired by human object recognition, which has been shown to rely on contextual information as opposed to classical appearance based methods. While this concept may not be new, this research sought to develop generic methods that leveraged recent developments in cognitive systems research, and more specifically large scale ontologies or knowledge bases. The results of the research have shown that context-based methods, supported by an ontology, can increase recognition rates versus classical appearance based methods. These methods have the potential to automate many complex object recognition tasks, aerial imagery analysis being one of them, that currently require human analysis.

Modeling and Analysis of Space Based Transceivers

This paper presents the tool chain, methodology, and initial results of a study to provide a thorough, objective, and quantitative analysis of the design alternatives for space software defined radio (SDR) transceivers. The approach taken was to develop a set of models and tools for describing communications requirements, the algorithm resource requirements, the available hardware, and the alternative software architectures, and generate analysis data necessary to compare alternative designs. The space transceiver analysis tool (STAT) was developed to help users identify and select representative designs, calculate the analysis data, and perform a comparative analysis of the representative designs. The tool allows the design space to be searched quickly while permitting incremental refinement in regions of higher payoff