Shanhe Yi, Computer Science - Dissertation proposal
Not only the rapid shifting from hand-hold mobile devices to ubiquitously connected smart things, but also the emerging new type of applications (e.g., AR/VR, real-time video analytics, cognitive tasks, etc.) and various mobile backends (e.g., cloud computing, edge computing, etc.) have brought both challenges and opportunities. We identify the challenges as: 1) new interface design challenge since the re-purposed legacy mobile system fails to consider adapting user interfaces to the way the mobile devices are used; 2) user authentication challenge that expects authentication schemes guarantee both security and convenience; and 3) mobile backend infrastructure challenge which requires integration of emerging mobile backend infrastructures to support new mobile applications. In this dissertation proposal, we present three research projects to address these challenges.
First, we find that current interfaces of smart glasses are difficult to use, error-prone, and provide insecure authentication. We present GlassGesture, a system that improves Google Glass through a head gesture user interface with gesture recognition and authentication. We speedup the recognition by employing a novel similarity search scheme. We improve the authentication performance by applying new features based on peak analyses of head movement in an ensemble learning method. GlassGesture achieves 96% gesture recognition accuracy. GlassGesture accepts authorized users in near 92% of trials, and reject attackers in near 99% of trials.
Next, we investigate the token-based authentication and the proper authentication channel between a smartphone and a paired smartwatch. We design and implement WearLock, a system that uses acoustic tones as tokens to automate the smartphone unlocking securely. We build sub-channel selection and adaptive modulation in the acoustic modem to maximize unlocking success rate against ambient noise only when those two devices are nearby. We reduce the unlock frequency by leveraging the motion sensor on the smartwatch. We also offload computational tasks from smartwatch to smartphone for quick response and energy saving. WearLock automates the unlocking with at least 18% speedup compared to traditional manual personal identification numbers (PINs) entry, and achieves a low bit error rate (BER) as 8%.
In regard to leverage emerging mobile backend infrastructures, we consider low-latency video analytics on mobile devices. We propose LAVEA, a system built on top of an edge computing platform, which offloads computation from mobile clients to edge nodes, and collaborates nearby edge nodes, to accomplish timely tasks with intensive computation at places closer to users.We outline the main idea and present the on-going work of LAVEA with a novel application.
Shanhe Yi is a Ph.D. candidate of Computer Science at William & Mary, advised by Dr. Qun Li. His research interests include mobile computing and edge computing. He received his B.Eng. in Communication Engineering and M.S. degree in Communication and Information System from Huazhong University of Science and Technology.