Research Experiences for Undergraduates (REU)
Each project will begin with an in−depth study of the scientific concepts that underlie each project. Students will also have the opportunity to observe the more theoretical activities of their graduate mentors and to see the results of their work in the broader context of the research objectives. Each student will also be required to work with graduate students in publication activities on the results of their work.
Adaptive Event Stream Processing for Security in SmartGrid (Dr. Susan Urban)
Software systems for contemporary applications are becoming increasingly complex, demanding the use of active, reactive, and proactive behavior to support context−aware and adaptive execution environments. To support these new application requirements, this research is transforming event stream processing into a dynamic and adaptive process through the integration of event detection with machine learning. This new approach to event processing involves the development of an adaptive environment for event detection that combines event query processing and logical inference with probability measures that can be used to detect well−known event patterns, to evolve existing event patterns, and to learn new and meaningful event patterns. This research is being conducted in support of two separate NSF grants for which S. Urban is a co−PI. One grant (EECS−1040161) is an MRI award to an interdisciplinary group of faculty to develop a real−time simulator for SmartGrid systems integrated with distributed renewal energy sources, with a focus on cybersecurity. The other grant (DUE−1241735) is through the NSF SFS (Scholarship for Service) program to develop an interdisciplinary cybersecurity education program for protecting critical infrastructure. REU student projects will focus on developing testbeds, developing and testing event detection patterns for intrusion detection in the Smart Grid, and performing some of the statistical analysis and data mining that is needed to support the learning of new event patterns. Participants in the previous NSF REU Site program have already established a preliminary Zigbee home area network testbed.
Software Specifications for Cybersecurity (Dr. Joseph Urban)
The Descartes specification language research effort is one part of an overall software engineering research program in software requirements, specification, process, and environments. The Descartes specification language effort is based on a solid foundation of research that has resulted in graduate students who completed degrees and the involvement of three REU site project students and two REU supplement students. The research effort will focus on automated software specification generation from expected input and corresponding output that is a direct extension of earlier research on automated test data generation from Descartes specifications. There are extensions to the language for real−time, object−oriented, and intelligent agent software development that will support an effort on executable software specifications as formal methods for information assurance, investigated in the context of the MRI and SFS grants mentioned above for Susan Urban. Dr. J. Urban is the P.I. of the NSF SFS project.
Autonomy in Human−Robot Collaboration (Dr. Mohan Sridharan)
Although mobile robots are increasingly being used in real−world applications, unforeseen changes in real−world domains frequently make it difficult for robots to operate without any human supervision. Towards this objective, Dr. Sridharan is developing an integrated framework that jointly addresses the associated learning, adaptation and collaboration challenges that enable robots to collaborate robustly with humans. The framework will enable robots to autonomously:
- Learn models of domain objects and adapt to unforeseen changes
- Adapt learning, sensing and processing to the task at hand
- Learn associations between multimodal object descriptions to pose verbal queries to humans
- Merge high−level human feedback with the information extracted from sensory cues.
Undergraduate students will be able to participate in the design, implementation and evaluation of these algorithms in simulation, and on humanoid, wheeled and aerial robots. Key application domains include surveillance, assistive care and reconnaissance. Students can also evaluate these algorithms in the international robot soccer competitions, where a team of humanoid robots plays a competitive game of soccer on indoor soccer fields.
Modeling Uncertainty in Climate Downscaling and Irrigation Management (Dr. Mohan Sridharan)
Climate forecasts influence policies and planning in fields such as agriculture, ecological preservation and resource management. Although sophisticated global climate models can predict large scale weather patterns, they cannot make accurate regional weather predictions since they do not account for local geographic variations such as mountains and lakes. In collaboration with the Climate Science Center at Texas Tech University and the Geophysical Fluid Dynamics Labratory/Princeton, Dr. Sridharan is developing deep architectures to learn the relationships between global models and regional observations, thus making accurate regional predictions. Similar challenges exist in agricultural irrigation management in arid and semi−arid regions, where crop water demand exceeds rainfall. For instance, daily grass or alfalfa reference evapotranspiration (ET) values are used to estimate crop water demand. Inaccurate reference ET estimates can impact irrigation costs and the demands on U.S. freshwater resources. In collaboration with United States Department of Agriculture−Agricultural Research Service and Texas A&M University AgriLife Research, Dr. Sridharan is developing machine learning algorithms that use historical weather observations from non−ET stations to accurately predict reference ET values, significantly reducing resource wastage. Undergraduate students will thus help design machine learning algorithms for critical challenges in real-world application domains.
DOROTHY: Design of Robot Oriented Thinking to Help Youth (Dr. Mohan Sridharan, Dr. Joseph Urban, Dr. Susan Urban)
Three-dimensional graphical programming environments such as Alice have been developed as an effective way to stimulate interest in computing by teaching students how to program. Students have also embraced robotics as a means to learn computing because robots illustrate practical applications of computing. In collaboration with five REU Site project students, we have developed a novel educational tool known as DOROTHY that integrates Alice with autonomous robots, with bidirectional communication between the graphical interface and robots. Students without any prior programming experience can create graphical routines in virtual worlds using syntax that is easy to learn. The tool automatically converts these routines to programs for synchronous and asynchronous execution (and adaptive behavior) on multiple different robot platforms in the real−world. Furthermore, we have developed a curriculum that can be used with this tool to teach core concepts of computing, concurrent execution and real−world sensing to school students. Undergraduate students involved in this project will enhance the tool and the curriculum, and also participate in teaching the material to local school students and teachers.
Fault−Tolerant Robot Applications (Dr. Michael (Eonsuk) Shin)
This project involves constructing reliable component−based robot applications. As robots are getting used in various application areas, the applications are becoming increasingly more complex. Intelligent and autonomous robots, such as unmanned ground vehicles, perform various tasks in an uncertain and dynamic environment. These robots may cause disastrous events if they encounter failures such as sensor or actuator failures, or periodic real−time data handling failures. Such failures should be detected and managed as soon as possible. Students involved in this project will investigate how to detect faults or failures in component-based robot applications and how to handle those failures.
Secure Software Engineering (Dr. Michael (Eonsuk) Shin)
This project involves constructing secure software by means of secure software engineering, which combines the principles of security and software engineering. Security has become an important requirement for many applications and needs to be considered during all phases of software development. The security features required by secure systems are confidentiality, integrity, non-repudiation, access control, authentication, and availability. To develop a secure software system, some or all these security features should be included into the system via software requirements specification, requirements analysis, software architectural and detailed design, implementation, and even verification and validation. This project will investigate how to combine the security features with software engineering at each development phase. Possible research topics would be threat analysis techniques, security requirements modeling, secure connectors for secure software architectures, secure detailed design, mapping from design to secure code, development of test cases for secure systems, and software intrusion detection.
Multi-Order Data Deduplication with GPGPUs for Data-Intensive Computing (Dr. Yong Chen)
Data deduplication has been generally recognized as a critical technique that reduces the data volume to be transferred over the interconnection and to be stored on storage devices in data-intensive high-performance computing, Cloud computing, and Big Data computing. Current data deduplication solutions, however, suffer costly byte-by-byte comparisons in the cases of hash collisions. This project investigates a Multi-Order Data Deduplication with GPGPUs method (MODD in short) to address the costly comparison issue in conventional data deduplication solutions. It reduces hash collisions and thus the need and the cost of byte-by-byte comparisons. It further trades computation capability to data access capability (further reduced data movement), similar as the idea of the original data deduplication technique. The MODD method holds a promise and can be widely applicable to high-performance computing, Cloud computing, and Big Data computing.
Fusion Active Storage for Write-intensive Big Data Applications (Dr. Yong Chen)
Many high-performance computing applications have become highly data intensive due to the substantial increase of both simulation data generated from scientific computing models and instrument data collected from increasingly large-scale sensors and instruments. These applications transfer large amounts of data between compute nodes and storage nodes, which is a costly and bandwidth consuming process. The data movement often dominates the applications¹ run time. This project investigates a new Fusion Active Storage System (FASS) to address the data movement bottleneck issue for write-intensive big data applications. The FASS enables a paradigm that moves write-intensive computations to storage nodes, generates and writes data in place to storage devices. It moves computations to data and avoids the data movement bottleneck on the data path. The FASS has a significant advantage of minimizing data movements and can have a profound impact on big data applications.