Pervasive Computing Laboratory

People

Brahim Bansaou Michael Brown Gary Chan
Dik Lee Bo Li Qiong Luo
Lionel Ni Long Quan Qiang Yang

Location

Projects

WAMS: A Context-Aware Map Server (Dik Lee)

This project develops a map servers to deliver map and context-sensitive information to the user over a wireless network. It encompasses several research and developmental issues:

PDA-based Web Browsing (Dik Lee)

This project takes a proxy approach in adapting web contents for PDA browsing. The user enters a URL and the proxy fetches it from the web, analyzes its structure, and present a navigation structure for the user to explore the page.

References

COPS: Collaborative Pervasive Surveillance (Gary Chan)

Traditional closed-circuit video surveillance systems are based on cabling and hence is not pervasive and low-cost. With users being increasingly mobile, we develop a surveillance system named COPS which allows users to monitor a place not only over the Internet, but also over the Wireless LAN and cellular networks using their cellular phones or PDAs. Furthermore, users would be automatically alarmed if suspicious activities are detected by the computer. Video storage, indexing, and retrieval will also be offered for later playback and searching. In this system, users can conveniently, collaboratively, and securely monitor a place at any time, from anywhere.

Cyberclassroom: Internet-based interactive distance-learning (Gary Chan)

We develop a large-scale live interactive distance-learning project termed "cyberclassroom" where students anywhere can attend lectures using wired and wireless medium with their PCs, PDAs, or phones. Instructors may poll students and students may raise questions at any time in the system. The students may also record lectures for later playback if they want.

Large-scale application-level multicast (Gary Chan)

Multicast is a fundamental and important mechanism to deliver scalable multimedia services. Unfortunately the Internet nowadays still cannot support global multicast. In this part of the project, we develop efficient networking techniques to enable global multicast. The technique, termed application-levle multicast (ALM), which takes advantage of local multicast-capability of IP networks while overcoming the limitation in multicast routers. Our scheme connects the multicast-capable "islands" together by means of overlay links and hence is termed "Island Multicast," which is able to support potentially hundreds to thousands of users concurrently.

Ubiquitous wireless access (Gary Chan)

Data services are increasingly delivered over wireless medium. Currently, we are developing techniques to integrate Wireless LAN (WLAN) infrastructure mode, ad hoc mode, and GPRS. Users would be automatically connected to either one of these modes to maximize their performance in terms of speed, cost, etc. By localizing user traffic, the overall system throughput can be substantially increased.

Location-based Plan Recognition for Pervasive Computing (Qiang Yang)

In this project, we develop a novel plan-recognition technique to recognize a mobile user's plans based on location sensors and user-behavior models using statistical learning, planning and constraint reasoning. When a user's location-sequences are observed, the plan-recognition algorithm predicts the users' next actions and goals using a collection of information sources such as the sensory data from GPS or infrared sensors. Our approach is to learn a user's behavior patterns that describe events, time and locations at different levels of abstraction. The learned model can be used to predict future goals and plans as well as the user's future locations, providing a basis for many location-based services such as location-dependent content delivery, intelligent bandwidth allocation, content caching and pre-fetching for M-commerce. Our research will focus on how to learn the high-level behavior models and a hierarchical plan library from the past location sequences and knowledge about the actions. A novelty of the work stems from the integrated statistical and logical representation and reasoning methods for locations, time and plans. Another novelty is the ability to directly recognize a user's plans from location-based sensors, making it possible to infer high-level knowledge in pervasive computing applications.

Transparent Roaming for Mobile Users (Bo Li)

This project develops enabling technologies to allow mobile users to roam across different platforms, such as WLAN, WWAN and ad hoc networks. The key is to provide transparent roaming support, which integrates physical, link and networking layer functionalities.

Overlay Platform for Real-Time Peer-to-Peer Applications (Bo Li)

This projects aims to build a overlay network prototype to enable more efficient communications for real-time P2P applications. There are three fundamental issues that will be investigated: 1) how to support scalable routing; 2) how to deal with the highly dynamic nature of the P2P applications; and 3) How to provide minimum delay. This is a joint project with Microsoft Research.

Aorta: Action-Oriented Pervasive Query Processing (Qiong Luo)

A key insight in this research is that pervasive computing applications require integrated data acquisition, communication and actions across heterogeneous devices. For example, when a motion sensor detects someone entering a restricted area at an unexpected time, the surveillance application will control a camera to capture the intruder's image and to alert the human guard via the phone. Therefore, we propose an action-oriented pervasive query processing software framework to extract real-time data from a network of heterogeneous devices. Furthermore, this framework will enable an application to communicate and to take actions across the devices.

We propose to study three research issues in action-oriented pervasive query processing: (1) a declarative interface for applications to specify their queries and actions, (2) a uniform data acquisition and transmission layer across devices, and (3) distributed execution techniques for queries involving actions. Our framework will enable a wide range of useful applications over a network of heterogeneous devices.