A Prototyping Environment for providing Real-Time Calibration & Dynamic Resolution Controlled Localization Information to enable Ubiquitous Computing (Joseph Ng, et al.)
Description:
This project is about how mobile and wireless technologies can play a role in ubiquitous computing systems. We propose to design and construct a prototyping environment to build systems and applications related to ubiquitous healthcare systems.
Instead of reinventing the wheels, we make use of the smartphone, off the shelf components, and existing technologies in ubiquitous computing (i.e. wireless & mobile positioning technology, and data acquisition techniques and processing via sensors from the smartphone) and together with data management (i.e. data capturing, data storage and retrieval, data/signal processing) to build middleware, APIs, and tools for the development of systems and applications for ubiquitous healthcare systems.
Two major tasks are identified: 1) To build a test bed to provide the infrastructure and architectural support for realizing ubiquitous computing; 2) To design and construct mobile applications, tools and services within the test bed for users to experience the benefits and practicality of the system; We use scenarios to illustrate how mobile / wireless and sensor technologies can enable ubiquitous healthcare. Some of these examples are: User Location Tracking; Well-being Data Acquisition & Analysis; Fall & Gesture Detection and Behavior Tracking.
Hence, the goal of the proposed project is to construct a prototyping environment to build systems and applications for providing ubiquitous healthcare services. As a start, the development of this project is geared toward an elderly centre setting. Systems, applications and services will be designed to serve the elderlies and demonstrate how mobile / wireless and sensor technologies can improve quality of living and enable ubiquitous healthcare within the community.
Project Objectives:
Grant Support:
Research Findings:
The Principle Investigator of this research study had been heavily engaged in four research projects related to ubiquitous computing funded by RGC and the Innovation Technology Commission through the ITF Funding schemes with a total project amount of $2.5M. The MLES, ILAS, and Smart AP projects funded by ITC and the Research Centre for Ubiquitous Computing (RCUC) funded by RGC had laid solid ground work on localization based on the wireless infrastructure.
Reported in [1], it describes a series of experiments that investigate different methods in capturing the signal strength characteristics to the signal database during the training phase of the fingerprint approach. The experiment also covers the Wi-Fi signal propagation in a long range and in a semi-outdoor environment. Furthermore, although the fingerprint approach provides good location estimation, the cost for training and maintaining the signal database is huge. The last experiment in this series is to construct and adopt the Center of Gravity algorithm for location estimation. Location estimation is conducted in real-time and does not rely on a trained signal database. It was found that location estimation is good when the users are around the centroid of the convex hull as defined by the access points of the WLAN. However, the algorithm loses its accuracy when users move to the rim the convex hull.
In [2], a method called “Aggregated Signal Layout” was proposed to find out the maximum likelihood of a user’s location based on the signal strength received. Instead of using the fingerprint approach which is costly for training and maintaining the signal database, the construct of signal layout maps for each AP is less costly and by stacking up the signal readings from each signal layout map, the user’s location can be estimated. Experiment results have shown that location accuracy that is comparable to the fingerprint approach can be obtained while the cost of maintenance of the signal database and the problem arising from the tear-down and the replacement or the installation of a new access point can be resolved. The overall architecture of the location estimation system was fully described in [3].
Since the Center of Gravity method produces very good estimation near the centroid of the convex hull, we, therefore, use the Center of Gravity method as an Algorithm Selector for location estimation. We first use Center of Gravity to estimate the user’s position. If the first estimation is near the centroid of the convex hull, it should be accurate enough for positioning purpose. Otherwise, we use other algorithms (e.g. fingerprint, or aggregated signal layout,) for location estimation near the rim of the convex hull [4, 5].
The PI and his collaborators also had been working on personalized tracking systems taking the fusion approach in merging different kinds of positioning technologies (e.g. NFC and ZigBee) with the Wi-Fi networks. Furthermore, the PI had also design and construct a SmartHealth series of experiments in making use of the smartphone’s computation power and sensor technologies (e.g. Accelerometer, Gyroscope, Magnetometer, Microphone) to capture and analyze pervasive data changes, user gestures, detection and falls, and user movement behavior.
Related Publication:
For further information on this research topic, please contact Prof. Joseph Ng.
|