Welcome to My Homepage

 

gutao

name

Assistant Professor
Department of Mathematics and Computer Science
University of Southern Denmark
 
Address: Campusvej 55, DK-5230 Odense M, Denmark
Office: Ø17-603a-2 Tel: +(45) 6550 2309
Email:

[Publications] [Services] [Grants] [Teaching] [Award] [Lab] [Group] [My family]

Highlights:  

Short Biography

Research Interests

Research Lab

Teaching

Award

Professional Services

   Journal Editorial Board Membership:

   Conference Steering Committee:

   Conference Organizer:

   Conference TPC Membership:

Research Grants

Publications [citations: 2100+]

   Recent Journals:

   Recent Conferences:

Talks

Group Members

   Current Members:

   Former Members:

Selected Research Projects

fig2

Multi-user Activity Recognition in Wireless Body Sensor Networks

Pervasive computing is continuously evolved into the next step such as cyber-physical systems; human has been again placed into the central point of a system. Understanding human behaviors is particularly important in bridging and integrating the cyber and physical spaces, especially in a multi-user scenario. This project aims to investigate the fundamental problem of recognizing activities for multiple users using Wireless Body Sensor Networks (WBSNs) in a home environment. We design a multimodal WBSN to capture observations for multiple users, and propose a novel pattern mining approach to recognize multi-user activities. To the best of our knowledge, this is the first reported work on multi-user activity recognition in WBSNs.

This work has been reported in MobiQuitous 2009 and TMC. The dataset and code are available here (dataset and code).

 

fig1

Single-user Activity Recognition in Wireless Body Sensor Networks

Recognizing human activities using WBSNs is particularly challenging because human activities are often performed in not only a simple (i.e., sequential), but also a complex (i.e., interleaved and concurrent) manner in real life. In this paper, we develop a WBSN, and conduct trace collection in a complex, real-world environment. We then design our algorithms to recognize sequential, interleaved and concurrent activities in a unified framework. As far as we know, this is the first reported work of a pattern mining approach to complex activity recognition using WBSNs.

This work has been reported in PerCom 2009 and TKDE. The dataset and code are available here (dataset and code).

 

fig3

Context-aware Infrastructure for Pervasive Elderly Homecare

Context-awareness is a key enabler for the adaptation of operations or behaviours of pervasive applications without or little explicit human intervention as possible. This project proposes a horizontal service-oriented infrastructure for the fusion of cyber and physical spaces and facilitating intelligent processing. Our objective is to develop the techniques and algorithms for context-awareness, as well as to demonstrate the feasibility of the infrastructure for supporting various applications over multiple context spaces consisting of sensors, devices, human users and applications.

This work has been reported in several conference and journal papers including UbiComp 2008 and JSAC.

 

fig4

Service-Oriented Context-Aware Middleware

This project aims to develop a distributed infrastructure for designing reliable, scalable and self-organized pervasive computing applications. The infrastructure addresses several key issues commonly exist in pervasive computing systems such as context data modeling, peer-to-peer searching, and context reasoning. This work is part of my PhD thesis.

This work has been reported in several conference and journal papers. The source code is available here, socam and scs.

 

Locations of visitors to this page

Free counter and web stats