Yunhao Liu, Ph.D. - US grants

The requirement on timely processing and analysis of huge volumes of data generated by various applications in our daily lives has driven an steady but urgent need to speed up the computing power. However, due to the physical limits in transistor scaling and the cross-core interference in multicore systems, the computer processor design for large-scale parallel computing is facing its limits. A new network-level parallel computing architecture and innovative performance optimization algorithms developed in this project can free the speedup of computing power and completely break the barriers in computing processor hardware design. The new technologies resulted from this project can be widely used in many parallel computing related applications for timely analysis of big data and secure data storage. Moreover, by integrating the technological advances resulting from this project into the undergraduate/graduate curricula and outreach activities, this project has significant impacts on the training of a highly-skilled and diverse workforce for high-performance computing.

The goal of this project is to achieve a theoretically-optimal speedup of computing power by leveraging the computing capability of the edge and cloud computing resources in the computing network, so as to enable efficient, scalable and secure parallel computing at the network level. More specifically, (i) this project develops new network-level parallel computing architectures and innovative performance optimization algorithms that can achieve a linear speedup of computing power by leveraging the computing capability of the edge nodes and cloud computing servers in the network; (ii) this project breaks new ground in developing secure parallel cloud computing and data storage schemes. The new techniques make it possible to implement efficient network-level parallel computing schemes without compromising the security requirements, and hence pave the path for the wide adoption of network-level parallel computing in future big-data applications.

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.

This project aims to develop non-intrusive and universal vibration sensing schemes that can detect the abnormal vibrations of a running machine. Towards this goal, the researchers propose a system that first uses the backscatter signals in commercial off the shelf RFID systems to accurately measure machine vibrations, and then uses machine learning and signal processing techniques to detect abnormal machine vibration patterns so that machine operators can be alerted to take actions before the machine fails. This project represents an emerging space driving new CPS and Internet of Things concepts for machinery safety. It can be used for the prognostic monitoring of not only indoor machines, but also outdoor appliances and civil infrastructures, such as drilling system monitoring, pumping system monitoring, pipeline system monitoring, and bridge monitoring. The proposed system is expected to impact manufacturing and economy. This project will bridge the communities between Computer Science and Mechanical Engineering; and foster interaction and communication among them. It will also facilitate the effort of the researchers on attracting and mentoring undergraduate students and underrepresented graduate students in research. Furthermore, the researchers will integrate the research results from this project into both undergraduate and graduate curricula.

This project has two key technical objectives: to develop vibration measurement schemes using RFID systems and to develop abnormal vibration pattern recognition schemes based on the measured vibration signals. For vibration sensing, the basic idea is to measure the machine vibrations through random and low-frequency readings of the tag using the RFID reader, where each reading is viewed as one sampling of the vibration. For abnormal vibration pattern detection, the basic idea is to build base line models based on the measured vibration readings and then classify real time vibration readings of a running machine as being either normal or abnormal. The proposed system would have several advantages over prior art in machine health monitoring , e.g., nonintrusive, inexpensive, accurate, and easily deployable including in non-line-of-sight scenarios.

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.

By moving computation to the edge of the Internet, the emerging edge computing model promises to reduce application response time, improve user experience, save bandwidth cost, and enhance data privacy. This project will establish a new edge-computing theory called coalescent computing, where each application program runs on an integrated virtual computing system, consisting of resources from both edge devices and the cloud, which jointly support application execution according to run-time conditions. The proposed research will establish a novel framework of edge-cloud resource sharing, explore various implementation mechanisms, deploy a hardware platform for experimental studies, develop software tools to transform existing applications to their coalescent-computing versions, and carry out two case studies. Its objective is to enable user applications to seamlessly overcome the resource limitation of edge devices, while keeping the benefits of low response time, communication reduction, and data privacy that edge computing promises.

This work addresses the pressing need for a generic theory of adaptive edge computing, with concrete implementation mechanisms that allow existing applications to take advantage of edge computing without complete re-design. The proposed coalescent-computing framework and its software tools offer a convenient way of transforming existing application code into distributed software with enhanced performance, response time, robustness and scalability, which all contribute to better user experience. The integrated theoretical and experimental research will advance our understanding of modern edge-computing systems. All software tools developed from this project will be made open-source to help stimulate other related research. The project also includes an education plan that introduces edge computing into undergraduate/graduate courses, recruits from under-represented student groups, and helps in club activities at the local high school.

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.

Emerging mobile applications are increasingly focused on technology interacting with and augmenting the real-world environment the user occupies. Augmented reality is a technology that places virtual objects on a user?s view of the real world with a wide range of applications such as navigation, gaming, and education. Augmented reality as a technology is inherently extremely computation-heavy, leading to latency, accuracy, and energy-consumption issues on resource-constrained smartphones. Image-recognitio- based augmented reality compounds this issue by requiring the computation of the entire image-recognition pipeline. In addition, mobile hardware is not designed with augmented reality and heavy image-based computations in mind. Mobile caching, multicores, GPUs and other mobile architectures are not being utilized to their full potential to help resolve the issues plaguing mobile augmented reality. This project explores methods to utilize the unique mobile architecture of off-the-shelf smartphones in new ways to realize augmented reality on a wide variety of mobile devices. Specifically, augmented reality has become an important tool for educators at all levels from K-12 all the way through collegiate and post-graduation education. This project will allow for this new educational technology to be more widely utilized in the world.

The objective of this project is to enable smartphones to support augmented reality via efficiently and seamlessly computing image-recognition and world-tracking tasks simultaneously, with three research components. (1) Investigate the foundational issues of smartphone-based augmented-reality through approximate-tracking to provide high-quality object tracking at reduced computational and energy loads. (2) Research software-defined caching techniques to utilize the unique nature of mobile augmented reality to provide caching specially designed for highly collaborative device-to-device augmented reality. (3) Explore hardware-based techniques relating to GPGPU computation and cache management to facilitate fast and efficient image-recognition and augmented-reality tasks.

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.