I am generally interested in mobile/edge AI systems, where I design and optimize networks and computer systems for supporting emerging (mobile) intelligent applications. Specific research topics include:
Here are some highlighted ongoing work:
DNNCare aims to enable automatic mobile/edge resource provisioning for deep learning based mobile applications (e.g., augmented reality, video analytics) through holistic resource management. The goals are resource efficiency, high throughput, low latency or SLA guarantee.
Griffin is a distributed storage service tailored for (collaborative) edge applications to share state/data. Griffin features designs that address challenges for the edge environment, which include expressive abstractions and consistency semantics and network-global monitoring and optimization for heterogeneity and mobility support.
INC proposes to use programmable switches to accelerate application-specific computations. We are exploring how to support modern workloads like machine learning with in-network computing and also provide abstractions to ease the use of in-network computing.
STEAM is a distributed back-pressure-based packet scheduler for service function chaining in packet processing networks, with the goal of high resource efficiency and low packet drop rates.
SmartEdge is an edge resource management/orchestration framework with algorithms that make informed decisions on resource allocation and service placement in edge computing with proven performance guarantees.
[ToN’20] [TMC’19] [INFOCOM’19] [INFOCOM’18] [JSAC’18] [ICDCS’17]
Microservice Autoscaling and Caching for Edge Computing
Funded by the German Research Foundation (DFG) within the CRC MAKI
SmartEdge: Concepts and Methods for Edge Computing
Funded by the German Research Foundation (DFG)
Edge Computing for the Internet of Things
Funded by TU Darmstadt (Athene Young Investigator Award)
@ 2020 Lin Wang. All rights reserved.