Fei Hu - US grants

Computer Science (31)

Pervasive Computing is becoming more and more prevalent in our society. Rapid developments in wireless technologies and Sensor-network-based smart spaces are creating an urgent need for well-trained engineers in Pervasive Computing. Unfortunately, most universities have only graduate-level Pervasive Computing courses, which may not be suitable for undergraduate learning. Rochester Institute of Technology (RIT) is currently undertaking an effort to enhance the pervasive computing skills of our undergraduate students in two programs, Computer Engineering (in College of Engineering) and Computer Science (in College of Science).

Intellectual Merit: This project is enhancing RIT undergraduate computing skills by adapting and implementing some of the exemplary Wireless & Mobile Computing lab environments at Benchmark Universities (Cornell, Rutgers and UIUC). RIT is establishing a Pervasive Computing Laboratory (PCL) that provides an integrated wireless-plus-wired networking platform, including current pervasive computing platforms such as Wireless LAN based on IEEE standard 802.11g, personal area networks based on Bluetooth technology and wireless sensor networks. The PCL is providing the Computer Engineering students with an understanding of computer networking issues on different mobile hardware platforms, and is providing the Computer Science students with an understanding of data management needs within pervasive computing environments. Three new courses (Pervasive Computing Architecture & Design, Principles of Wireless & Mobile Networks, and Data Management for Pervasive Computing) and a series of laboratory assignments are being developed to enhance the undergraduate curriculum of both majors.

Broader Impact: The PCL is providing an excellent opportunity for the students and faculty to foster inter-college cooperation between two of the traditionally well-known undergraduate programs at RIT. By leveraging the multi-disciplinary character of RIT, the PCL is enhancing the diversity of pervasive computing skills in our nations' high-technology workforce since the students will be engaged in exploring several pervasive computing environments for monitoring/analyzing the physical world. In addition, RIT is using the PCL equipment and basic materials to train high school students through workshops.

Tele-healthcare could largely reduce national healthcare cost through remote self-managed patient monitoring. Cardiac Sensor Networks (CSNs) could be used to deploy such a system. Moreover, the integration of RFID into CSN could play an important role for elder healthcare because RFID could be used to monitor elders' medicine taking behaviors. On the other hand, the disclosure of RFID information during RFID tag-to-reader communications can cause the violation of patients' privacy. This research aims to achieve trustworthiness in a practical RFID-assisted CSN platform. This project will make the following three contributions: (1) Error-resistant ECG transmission: A trustworthy CSN should be able to overcome the impacts of radio interference and propagation distortions that can lead to frequent ECG transmission errors. This research will use the receiver-only local ECG processing to overcome ECG errors/loss. It will conduct a comparative study of two promising anti-interference methods, Kalman Filter and Tempo-spatial Regression. (2) Low-overhead RFID security: This research will design a lightweight RFID security scheme based on Linear Congruential Generator (LCG) and a mutual authentication scheme. The RFID security will be tested in our current CSN hardware platform. (3) CSN temporal accountability: In ECG sensor applications, all ECG anomaly detections depend on accountable time interval analysis between different ECG signal segments. This research will achieve CSN temporal accountability through the following two technical approaches: i) design a receiver-only clock uncertainty prediction model to avoid wireless communication overhead; and ii) design a history-aware, reputation based trust model to capture the evolutionary timing attacks.

EMT: Collaborative Research: Primate-inspired Heterogeneous Mobile and Static Sensor Networks

Although previous bio-inspired models have concentrated on invertebrates (such as ants), mammals such as primates with higher cognitive function are valuable for modeling the increasingly complex problems in engineering. Understanding primates? social and communication systems, and applying what is learned from them to engineering domains is likely to inspire solutions to a number of problems.
This research involves studying and modeling modes of group behavior and communication of coppery titi monkeys, rhesus macaques, and other primate models, and applying what the investigators learn to the distributed control of heterogeneous mobile and static sensor networks. The investigators will model the social and communication behavior of these primates, which will provide biological inspiration for solving problems in communication and networking. The phases of this research include: 1) identification, interpretation, and translation of primate behavioral models, 2) assessment of the effectiveness of small and large group formations based on primate grouping models in heterogeneous mobile and static sensor networks, 3) development of bio-inspired message-based communications, and 4) development of bio-inspired behavior-based communications. This research aims to achieve a deeper understanding of effectiveness of bio-inspired communications and networking by studying primates, and to establish interdisciplinary research and education in the fields of biological modeling, sensor networking, and robots control.

Although high-cost, data-intensive multi-camera systems have been widely used for mobile human tracking and recognition, the pyroelectric infrared (PIR) sensor has a variety of advantages including dramatically low costs, chemical stability, high sensitivity to human body thermal variation, and extremely low sensory data throughput.

This project implements an Intelligent Compressive Multi-Walker Recognition and Tracking (iSMART) testbed based on PIR Sensor Networks (PSN). The novelties of iSMART include three aspects: (1) Context-aware region-of-interest (RoI) exploration to achieve an inherent tradeoff between area of sensor coverage and degree of information acquisition resolution. This research uses strict mathematical models to measure RoI context. (2) Decentralized inference / learning for in-network intelligence. This project develops a belief-propagation-based distributed inference scheme with data-to-object association for continuous tracking and recognition of multiple walkers. It uses orthogonal-projection-based distributed learning for sensor calibration and feature model training. (3) Networked, compressive sampling structures and sensing protocols. This project extends the latest progress in compressive and multiplex sensing theories to guide the design of novel networked sensor receiver pattern geometries and decentralized sensing protocols.

The above research efforts will lead to a novel low-cost, high fidelity wireless distributed sensing system for multiple walker recognition and tracking. As an alternative to video camera systems, iSMART systems can be widely deployed to automatically monitor airports, customs / harbors, and other critical national infrastructures. This project will also generate interesting hands-on labs on intelligent sensor / sensor networks and class projects for both undergraduate and graduate students.

This project is developing a set of learning materials and laboratory modules in tele-healthcare to be used in undergraduate electrical and computer engineering curricula. The materials are organized into five project-lab trees and ten topic-subject trees that are self-contained and that may be used in any order. Specific topics covered by the new laboratory materials include design project in cardiac monitoring, mental health monitoring, medical security and long-distance medical transmission. The learning materials are being developed using a four-dimensional pedagogy that engages students with different learning preferences and allows for asynchronous delivery. The project includes rigorous formative and summative evaluation plans with both qualitative and quantitative components coordinated by an experienced independent evaluator. The evaluation plan is designed to establish the quality of the materials and to ensure the project goals are met. The projects results are being disseminated through conferences and journal publications and through direct contact with four other institutions of higher education with diverse student bodies. The materials developed in this project are being placed in the NSF sponsored National Science Digital Library.

Behavioral biometrics, such as gait, walking trajectory, and speech rhythm, are advantageous in their tolerance of low sensory resolution, long sensing distance, and poor environmental conditions. Many applications in security and tele-healthcare require tracking and identification of individuals based on their behavioral traits. Conventional approaches typically use video cameras, which could incur high data throughput and computational costs.
This project aims to develop a research infrastructure that enables tracking and identification of individuals? behavioral biometrics through thermal, pressure, photonic, acoustic, fiber-optic, and laser sensors. Besides low cost, those sensors can produce sparse behavioral biometric data that can be analyzed quickly. To achieve this goal, we will develop: (1) a set of tools for novel sampling geometries, cognitive sensing protocols, biometric feature modeling and distributed information aggregation; (2) a multi-agent architecture for system operation and adaptation that enables heterogeneous sensors to achieve consensus in behavior analysis; and (3) a set of programming and evaluation tools for system integration and improvement.
The developed computing research infrastructure will enable research activities about behavioral biometrics based on low-cost, low-power, distributed platforms with a variety of sensing modalities. The enabled distributed behavioral biometrics intelligence has great potential to cope with the technical challenges in behavioral biometrics caused by long distances and crowded scenes. The developed education materials will benefit a number of computing and engineering courses for the training of highly skilled workforce on surveillance and security monitoring. The outreach programs include summer student training, faculty workshops, and school visits.

This project establishes a multimedia-based virtual classroom with a virtual lab teaching assistant for the education of cyber physical system (CPS) security. Such a virtual classroom helps college students in resource-limited rural areas to learn the latest CPS security knowledge via an on-line peer-to-peer learning environment with other students from larger schools. This project includes three novel contributions: (1) all learning materials embrace an application-driven learning approach, with examples from diverse areas such as healthcare, renewable energy, and industrial controls used as the basis for CPS attack analysis; (2) with the help of a multimedia company, the project is building interesting virtual classroom lectures; and (3) to meet the open access lab requirements, the project is building interactive virtual lab helper software to enable remote students to conduct virtual hardware lab experiments and obtain help using multimedia tools. The design encourages innovative learning in several ways: developed labs require an iterative process with idea incubation to force students to follow a more mature creative design process; all labs intentionally include some ambiguity to encourage the search for multiple answers to a single problem; and the 3E (Explain-Exploit-Explore) based pedagogy is adopted for all CPS security labs/projects. The basic level labs emphasize concept explanations. The intermediate level senior projects require students to exploit previous knowledge to perform a multidisciplinary CPS security task. The advanced-level labs require independent exploration to reach creative solutions. The resulting teaching methodologies can be extended to other rural colleges and this project uses a proactive dissemination plan to achieve this aim.

Proposal #: 13-35263
PI(s): Hu, Fei
Qahouq, Jaber Abu; Brown, David A.; Gray, Jeffrey G.; Tsoupikova, Daria
Institution: University of Alabama ? Tuscaloosa
Title: MRI/Dev.: Robot-aided, Cognitive Virtual Rehabilitation for Automatic Physical Training of
Individuals with Disabilities (iRAPID)
Project Proposed:
This project, developing iRAPID, a robot-aided cognitive virtual rehabilitation instrument for automatic physical training of individuals with disabilities, integrates several hardware components (the KineAssist robot, programmable treadmill, biomarkers (sensors) and Xbox kinect sensors) to build an augmented virtual reality animation of the patient. A series of newly developed software tools will support virtual rehab research computations of body balancing and neuro-pattern changes during rehab.
The instrument will be designed to be suitable for three different cost/performance levels, with successive levels making use of more comprehensive sensors. Combining the sensors (e.g., functional NearInfraRed and EEG brain imaging) with physical rehab mechanisms (e.e., treadmill) make possible interesting research areas related to the effectiveness of physical rehabilitation training. Hence, iRAPID will be a cognitive, research-oriented rehab instrument with automatic, accurate rehab training progress computation. To enable the stated goals, the system should be capable of recording a wealth of sensor measurements. Advancing the next-generation rehab system, the work
- Adopts a hierarchical (3 layers), incremental (3 modes) development strategy,
- Supports computational rehabilitation research, and
- Adopts evolution-oriented software design.
Broader Impacts:
The instrumentation enables research an education of an exciting new field, Cognitive ElectroBiomedical Systems (CEBS), enables the training in CEEBS of two PhD students, and outreach to minorities. CEBS is a trans-disciplinary (TrD) field has distinguished features compared to multi-disciplinary (MuD) and inter-disciplinary (InD). The authors compare these fields to a cake where different ingredients are not easily distinguishable (TrD) giving a new format product, and a plate of salad that still has clear existence of different ingredients ((InD). The PhD training program will have a training structure included TrD/MuD/InD curriculum and service learning. The Director of Multicultural Engineering Program (MEP) will assist in involving underrepresented students with summer CAMP and K-12 activities. This development is likely to highly contribute within an EPSCoR jurisdiction.

The University of Alabama (UA) is designing and deploying a science DMZ herein referred to as UA SciNet to support the rapid expansion in research across multiple science and engineering disciplines. UA SciNet is based on the design patterns and best practices emerging from Internet2 and ESNet. UA SciNet is a dedicated, isolated network connecting to the UA wide-area network at the campus network border. It provides 40 Gbps connections between active data-intensive science areas and high performance computing and research storage resources. It provides an initial 400% bandwidth increase for research and lays the foundation for 100 Gbps connectivity to the wide area network via the UA System Regional Optical network and Southern Crossroads services. It includes a robust performance monitoring and problem resolution component using perfSONAR. It also uses wide-area remote DMA (RDMA) to achieve high-bandwidth remote access to UA Big Data resources.

The research impacted by the project includes hydro-meteorologic modeling, disaster event prediction and mitigation, human behavior, and particle physics, specifically leveraging the data-intensive operations at the recently completed National Water Center on the UA campus. This project also works in coordination with other initiatives both on campus and at the regional and state levels to ensure its effective integration and to ensure its benefits are fully leveraged. Further, the project brings unique education and training opportunities to undergraduate and graduate students, researchers and system and network professionals to the areas of advanced network infrastructure, data-intensive science, and performance monitoring and diagnostics.

Big data applications are widely utilized today in a variety of scientific and social studies. The International Data Corporation (IDC) reports that more than 50% of all big data business revenues originate in the United States. Big data requires "bigger responsibility" in terms of protecting the storage, sharing, and access of the privacy-sensitive data. In today's higher education, there are limited big data security (BigSec) educational activities that target comprehensive and profound understanding of BigSec attack models, as well as defense solutions, at the undergraduate and graduate levels. Some available BigSec training materials are mostly online, used for short-term training, and only cover general concepts from a business management viewpoint. The nation is currently experiencing a shortage of cybersecurity professionals, including the experts in big data security. The goal of this project from the University of Alabama Tuscaloosa is to utilize new pedagogy, called Captivology-Stimuli-based Learning (CAPITAL), for active education and training on big data security and privacy.

This project will make two significant contributions to enhance BigSec education. The first is the development of a new series of BigSec course materials, including an undergraduate course, CECS 400/500 Big Data Security, which targets fundamental knowledge of big data security issues, as well as a graduate course, CECS 600 Advanced Topics on Big Data Security and Privacy, which aims to enhance the cybersecurity research skills of graduate students. The second is the implementation of CAPITAL pedagogy in BigSec education. This includes implementation of the flipped classroom model, Virtual Reality based security games, Capstone design showcase, student cybersecurity design competitions, and patent-targeted thesis projects. This project will conduct the pioneering "active" education to train a BigSec-literate workforce. All the BigSec created education materials will be disseminated to other schools.

This project, developing one of the first Software Defined Network (SDN)-based underwater mobile testbeds to support the operation of marine robot fleets, aims to address a technological bottleneck, that of achieving integrated communications and navigation underwater. A fleet of Autonomous Surface Vehicles (ASV)s are directed to follow sampling Autonomous Underwater Vehicles (AUVs) to provide acoustic and Magnetic Induction (MI) communication over relatively short ranges. Launching the following effort thrusts: Acoustic & MI Communication; Control of ASVs & UAVs; SDN architecture; and Integration and evaluation.

The testbed will be designed to achieve cost-effectiveness, transferability, flexibility, and scalability and is expected to become a stable instrument that is accessible by multiple research communities that include ocean acoustics, communication and networking, robotics, oceanography and environmental sciences. The hybrid acoustic/MI communication will be used to achieve reliability and high data rates across the mobile network for ASVs and AUVs, while smooth autonomy of the fleet would be ensured by cooperative localization and real-time data transfer among the ASV-AUV pairs. The testbed is expected to enable various research directions, including underwater swarming, deep-learning-based underwater joint networking and navigation, and integrated oil spill responses.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

The broader impact/commercial potential of this PFI project is to benefit millions of post-stroke patients and other disabled people who need effective body rehabilitation, by developing an intelligent rehabilitation robot with adaptive upper trunk support and hologram-based virtual reality training scenes. The existing commercial rehabilitation products cannot efficiently support the upper torso of the patients with serious trunk disability, and do not have intelligent, fine-resolution robot control to train two legs with different impairment levels. An online rehabilitation training without frequent in-person visits to the office of the physical therapist (PT) would reduce the medical cost. This commercialization-oriented project will address those market needs. It will also develop and implement a new customer channel model called "8C + 5F", consisting of 8 communication channels with different types of customers and offering 5 technical facets of this platform. To enhance the academia-industry partnership, the team proposes an effective model called nexus organization (NEO), to achieve transdisciplinary, conflict-free collaborations between the university and the industry. To train future leaders in innovation and entrepreneurship, the team proposes a new education model called Innovation and Marketing-oriented Renaissance Foundry (IMRF) to prepare PhD students and postdoctoral researchers with strong entrepreneurship and innovation skills.

The proposed project aims to build a Holographic intelligent Rehabilitation robotic technology to train post-stroke patients. It has 3 marketing-oriented new designs: i) (Upper trunk exoskeleton with 3D linkage and elastic impedance control): A compact, light-weight and economic upper trunk exoskeleton will be developed for a robot system to improve the upper trunk stability control during rehabilitation. ii) (Intelligent platform control in each phase of intra-gait cycle to handle two-leg impairment asymmetry): Current treadmill control simply changes the speed/force of the treadmill in each gait cycle based on the user's speed or measured center of pressure. However, many patients have different impairment levels between his/her two legs. This PFI will use low-cost thermal/acoustic sensors with deep-learning-based 3D leg trajectory analysis to achieve fine-granularity 4-phase rehabilitation robot control. iii) (Self-engaging mixed reality rehabilitation environment based on holographic telemedicine): The team will extend their previously developed virtual reality (VR)-based design to a hologram-based mixed reality (MR) platform with the advanced telemedicine functions for virtual patient-to-PT or patient-to-patient co-rehab training.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.