General Areas
  • Architectures for networked devices, server systems, and data centers
  • Virtualized environment, cloud platform, and edge-oriented system
  • Design for availability, reliability, security, as well as disaster tolerance
  • Power, energy, thermal, carbon footprint, and capacity management
  • Evaluation, analysis, and characterization of computers and applications
Recent Focuses

Graph Processing Platform: Efficient programming model, system interface, and runtime support that allow one to acheive better design tradeoffs on heterogeneous graph computers.

Selected work: ICCD'17

Intelligent Edge Computing: Architecture design and system optimization for better utilizing ubiquitous, massive resrouces of smart devices that many future intelligent services rely on.

Selected work: ISCA'15, CSUR'18 , ICCD'18

Resilient Power Management: Protecting computing systems from malicious adversary who exploits the vulnerabilities within aggressive power provisioning and management schemes.

Selected work: DSN'15, ISCA'16, ICCD'18

Heterogeneous Infrastructure: Managing various heterogenous system components (both IT and non-IT equipment) to build more efficinet computing platforms for emering applications.

Selected work: ICS'15, HPCA'18

My Prior Work

Green Data Center Design: leveraging cross-layer power management solutions that span software, hardware, and renewable energy systems to improve the sustainability of data centers.

My dissertation research leverages promising renewable energy technologies and emerging hardware/software techniques to tackle the energy and environmental challenges faced by computers today and tomorrow. In the past, energy-efficient designs mainly emphasize reducing energy consumption or improving resource utilization. They are not aware of the behaviors of renewable power supply, and consequently overlook the interplay between power systems and computing systems. This work develops a holistic power management framework that enables highly efficient green energy integration in modern large data centers.