Youngmoon Lee

이영문 李永文

Assistant Professor, Robotics Department

Hanyang University, Korea

I am joining Robotics department at Hanyang University, Korea.

I was a Ph. D candidate in CSE department at University of Michigan. My advisor is Kang G. Shin and collaborator is Mosharaf Chowdhury. I graduated from Seoul National University in 2014.

I tackle battery and power/thermal challenges for robust IoT/AI systems on constrained embedded platforms. My research focuses on embedded and cyber-physical systems, and mobile/cloud computing.

I was born and raised in Korea. I served as a Korean International Volunteer in Rwanda speaking Kinyarwanda and French.

Research

Robust Embedded Autonomous/AI Systems

RT-TAS (EMSOFT'19) Vision systems co-schedules integrated CPUs and GPU by capturing platform’s thermal coupling; Chip overheating is mitigated by 6-12ºC
RT-TRM (EMSOFT'18) Capturing ambient temperature and optimizing power budget/resource utilization of control tasks; resource utilization is improved by 18.2%
eTEC (ISLPED'17) Co-optimizing system power and cooling power of thermoelectric cooler; reducing performance loss from 19.2% to 1.8% for mobile apps

Robust Embedded Battery Management

BPM Enabling mobile devices to predict usable battery power and regulate peak power consumption to avoid unexpected shutoff; operation time is extended by 2.03×
BaT (Ubicomp'20) pervasively provides ambient temperature estimation using phone batteries
EA-SoC (ICCPS'19) capturing real-time environment to estimate battery state-of-charge (SoC); reducing the end-of-discharge SoC from 40% to 3%.
Battery Scheduling (RTSS'16) co-scheduling multiple power-supply (battery, ultra-capacitance) and multiple power-demand operations; performance improves by 33%

Resilient Resource Disaggregation

Resilient Memory resilience mechanism for common memory disaggregation solutions handle remote failures, evictions, corruptions; improve app latency by 64×
Infiniswap (NSDI'17) Unifying cluster’s memory via RDMA with decentralized architecture to improve throughputs of big data applications 15× and latencies 61×