Jiadi Yu


Funded Research Projects

  • Research on Sensing Human Behavior Characteristics for User Identity Authentication Leveraging Millimeter-Wave Signals, National Natural Science Foundation of China (NSFC) (No.62172277), PI, Duration:2022-2025.

  • 面向HarmonyOS的细粒度人体特征感知技术, 华为终端有限公司, PI, Duration:2023-2025.

  • 自动驾驶试验场关键技术研究与应用验证, 上海市科学技术委员会(No.22Z510202440), co-PI, Duration:2022-2024.

  • 基于毫米波雷达的车辆驾驶舱泛在感知技术, 联想(北京)有限公司 Lenovo(202211SJTU02-LR004), PI, Duration:2023-2024.

  • Development of Wave powered Ocean Observation Profiler, Key Scientific Research Instrument Development of the National Natural Science Foundation of China (No.61827810), co-PI, Duration:2019-2023.

  • Research on Intelligent Urban Sanitation Operation Coupled System based on IoT,Big Data and Cloud Computing, National Key R&D Program of China (No.2018YFC1900700), co-PI, Duration:2018-2022.

  • Research on Sensing Human Vital Signs and Behaviors for User Interest Detection towards Physical Advertising leveraging mmWave Signals, CCF-Ant Group, PI, Duration:2021-2022.

  • Research on Sensing Vehicle Driving Behaviors based on Audio Signals, National Natural Science Foundation of China (NSFC) (No.61772338), PI, Duration:2018-2021.

  • Research on Secure Information Service and Big Data Application Key Technology, National Key R&D Program of China (No.2017YFC0803700), co-PI, Duration:2018-2020.

  • Research on Cooperative Active Safety in vehicular networks, International Cooperation and Exchange of the National Natural Science Foundation of China (No.61420106010), co-PI, Duration:2015-2019.

  • Research on Hybrid cloud Key Technology, National 863 Program, co-PI, Duration:2015-2017.

  • High Performance E-Trading Reporting, Collaborative Applied Research and Development between Morgan Stanley and University, PI, Duration:2015-2016.

  • Searchable Encryption in Cloud Computing, Doctor Program Grant from Ministry of Education, PI, Duration:2013-2015.

Current Research Projects

Internet of Things (IoT)

Human Reconstruction. Human reconstruction refers to the technology that utilizes the sensed data to reconstruct a user's whole body or body parts digitally, which plays an important role in human computer interaciton, virtual reality modeling, etc. In this project, we explore millimeter wave (mmWave) signals to sense and reconstruct users digitally, which provides a nonintrusive, privacy-preserving and illumination-robust human reconstruction. We first utilize mmWave signals to sense multiple users' daily activities and realize 3D posture reconstruction and tracking based on mmWave signals. Furthermore, we focus on facial expressions and realize more fine-grained 3D facial expression reconstruction based on mmWave signals. Related work is published is published in ACM MobiSys'22, ACM MobiSys'23.

Behaviors Recognition & User Identification. With the development of IoT, smart homes are gradually becoming a reality, which makes it necessary to effectively recognize behaviors and identify users. In this project, we explore widely-existed WiFi signals to recognize user behaviors and identify user identity, which provides interaction-based secure accesses for smart homes. We first design finger gesture-based continuous user identification using WiFi signals to simultaneously recognize fine-grained gestures and identify users’ identities. Then, we explore gesture-independent user identification, which leverages WiFi signals to identify users through arbitrary body gestures. Furthermore, we design a multi-user activity recognition and identification method based on WiFi signals. Related work is published is published in ACM MobiHoc’19, ACM MobiHoc'21, IEEE INFOCOM'22, IEEE TMC, IEEE TON.

Cyber Security and Privacy

Ubiquitous Side-Channel Attacks. In recent years, with the increasing occurrence of various eavesdropping attack events, side-channel attacks aiming at personal privacy have become one of the most worrying security issues. In this project, we demonstrate a possible side-channel attack which can infer keystrokes on QWERTY keyboard of touch screen through near-ultrasound signals. Such an attack employs the attenuation property of acoustic signals and further integrates tracking the finger movements between keystrokes to improve the inference accuracy. Moreover, we present a live voice eavesdropping method, which utilizes common glasses attached with a low-cost RFID tag to sense subtle facial speech dynamics for inferring possible voice contents. Relative studies are published in IEEE INFOCOM'19, ACM UbiComp'23, IEEE TMC, etc.

Behavior-based User Authentication. Daily behaviors in our life are probably extended to realize the user authentication to protect the security and privacy. In this project, we explore the individual uniqueness underlying the mouth movements to realize the user authentication with the Doppler profiles of acoustic signals leveraging deep learning techniques. Also, we exploit the physical characteristics of touching fingers on active vibration signals for realizing a behavior-irrelevant on-touch user authentication. Moreover, we explore to leverage Doppler shift of RF signals to sense human unique gait features for achieving environment-independent user authentication. Relative studies are published in ACM MobiCom'20, IEEE INFOCOM'19, ACM MobiHoc 2019, ACM UbiComp'20, IEEE SECON’22, IEEE/ACM ToN, etc.

Mobile Computing and Sensing

Vehicular Technology. As more vehicles take part in the transportation system in recent years, the development of vehicular technology becomes a corner stone to improve driving safety. In this project, we exploit two main areas, i.e., sensing vehicle dynamics and monitoring driving behaviors. For sensing vehicle dynamics, we exploit the motion sensors on mobile devices to sense the speed, lane condition and abnormal driving condition of vehicles. For monitoring driving behaviors, we explore acoustic signals to capture the abnormal behavior of drivers and track steering wheel angle of drivers in real time, and explore mmWave signals to realize heartbeat electrocardiograph revocery for monitoring driver's health condition. Related works are published in ACM SenSys’13, IEEE InfoCom’14,16,17,22, IEEE SECON’15,18, IEEE TMC.

Human-Machine Interactions. Human-computer interaction (HCI) is a multidisciplinary field of study focusing on the design of computer technology and, in particular, the interaction between humans (users) and computers/mobile devices/appliances. HIC significantly dominates the satisfactory of user experience while using specific systems. In this project, we explore the scrolling human-computer interactions during browsing to infer user interest about the browsing content through modeling the relationship between scrolling and user interest with naïve Bayesian and convolutional neural network techniques. Also, we investigate the relationship between user experience and color scheme during browsing and adapt it through sensing the ambient light. Relative studies are published in ACM UbiComp'19, ACM SenSys'15, IEEE TMC, etc.

Smart Healthcare

Vital Signals. Fine-grained breathing and heartbeat patterns are critical indicators of the wellbeing of people. However, existing studies on vital signal monitoring either require active user participation of wearing special sensors, or need relatively quiet environments. In this project, we leverage acoustic signal generated by mobile devices to estimate the fine-grained breathing waveform of users, which can separate breathing from other body movements and can be done in noisy environment like driving vehicles. Also, we exploit millimeter wave to detect the tiny body movement caused by heartbeats of users and further recover the fine-grained ECG of users. Related work is published in ACM MobiSys'19 and IEEE INFOCOM'22.

Some Past Projects

Cloud Security. Data sharing in the cloud, fuelled by favourable cloud technology trends, has emerging as a promising pattern in regard to enabling data more accessible to users in a convenient manner. To achieve data sharing, enterprises and customers in increasing numbers keep their data stored into cloud server. In this project, we focus on seeking a solution that allows secure and effective access to the cloud data and data privacy of searchable symmetric encryption (SSE) in cloud computing.

Cloud Resource Schedule. Benefiting from the pay-per-use pricing model of cloud computing, many companies migrate their services and applications from typical expensive infrastructures to the cloud. However, due to fluctuations in the workload of services and applications, making a cost-efficient VM configuration decision in the cloud remains a critical challenge. In this project, we first design a double auction mechanism to balance the profits of both cloud provider and users. Then, we model an optimization problem aiming to minimize the VM configuration cost under the constraint of migration delay. Taking advantages of Lyapunov optimization techniques, we propose a mix scale online algorithm which achieves more cost-efficiency than that of scale out strategy.