On-going Projects 在研项目
Here are some on-going projects of mine 📚
🏃♂️ AI-enhanced Human Activity Pattern Recongnition
IMU: uses IMU in wearbale devices (smart watch/smart phone) to capture the movement of human.
AI-enhanced model: uses AI-enhanced physical method to achive a good trade-off between accuracy and robustness.
Objective: develop a natural human device interaction method with current commerical wearable devices.
🎭 Privacy Lekage and Protection in Wireless Sensing
- Wireless Sensing: as recent advance in wireless sensing enable to capture diveristy human activity pattern through refected signal. It introduced new interface for privacy lekage attack.
- Side Channel Attack: through the wireless channel, the attackers can conduct a series of novel attacks, resulting in serious privacy lekage.
- Objective: investigate potential attacks through wireless channel and design the associate defense method for privacy protection.
🚘 Reliability Analysis of In-vehicle Network
- Vehicle Ethernet: Automotive Ethernet In-vehicle Network provides flexibility and freedom in the implementation of the in-vehicle EE architecture, which provides high bandwidth at low cost. However, the unrelibility performance of Ethernet communication become the major concerns of its adapation.
- DES: we uses the diseceret event simulation to build a digital replica of the in-vechicle network, which can be leveraged to efficiently test the network performance under varying network configurations and traffic models.
- DNC: we also leverage the network calculs to analyze the performance bound of the system, which helps to understand the wrost case delay bound and memory useage of the system.
- Objective: design an effective tool to valid and capture the potential risks in in-vehicle network communication.
📐 Adaptive Network Measurment with Sketch
- Limited RAM: as a switch chip needs to supports various of network functions, only small amount of RAM can be leveraged to measure the flow staitical.
- Long tail Flow Size Distribution: meanwhile, different flows usually has distinct features, and a small amount of elephant traffics accounts for most of network volume.
- Objective: design an effective sketch method to capture dynamic flows with distinct flow size