 Home |
| ALEIS Home
|
| Equipment Details |
| Multimedia |
| Download |
ALEIS: Assisted Living Embedded Infrastructure
ALEIS is an embedded
infrastucture targeted at Assisted Living Environments.
ALEIS uses RT-Chain and has been implemented on MICAz. It provides localization,
fall detection as well as voice communication. In this work, ALEIS demonstrates
how a sensor network enhanced with real-time communication can provide a reliable,
affordable, and non-intrusive solution for real-time monitoring
of Assisted Living facilities.
Using RT-Chain, ALEIS supports
real-time monitoring and emergency response. To validate its reliability,
ALEIS has been deployed for testing and demonstration
in Siebel Center. During the test, ALEIS was kept running for
about 96 hours. This experiment validated ALEIS's reliability and scalability.
In addition, ALEIS was deployed in an assisted living facility during a pilot study
(see Assisted Living project for more details).
Overview of Assisted Living Embedded Infrastructure
We propose Assisted Living Embedded Infrastructure (ALEIS) to
meet the goals of building an Assisted Living Environment that is reliable, non-intrusive,
extensible and affordable. ALEIS allows the elderly to live in a safer environment
with peace of mind by ensuring that their safety is monitored. This infrastructure
consists of the Monitoring System and the embedded hardware, MICAz Motes.
The Monitoring System is a software residing in a Personal Computer in the
assisted living environment itself. It can also be remotely accessed by healthcare
provider centers via the Internet. The Motes are further divided into
two types: Infrastructure Mote (IM) and Personal Monitor (PM). IMs are generally static
after being placed, while a PM is carried by an elderly. ALEIS has been implemented as
an extensible infrastructure that allows multiple different applications
of varying nature to be implemented on it.
Currently ALEIS provides
real-time emergency notification, voice communication, approximate localization and fall detection.
The following is a possible example scenario
of how these applications are useful: suppose an elderly wears the Personal Monitor
and is currently sitting in his room. The Monitoring System will reflect his location using a
layout. When he stands up, he feels dizzy due to postural hypotension and falls down. The
Personal Monitor will sound a beep and send an Emergency Notification to a base station.
The Monitoring System will record the event, and a supervising nurse can then choose to
open a voice channel to allow the elderly to request for help or simply to report that he is
safe.
We now provide more details about the three embedded applications developed on the MICAz motes
as well as the server side software that acts
as a control interface for users (or healthcare providers).
Testbed
A set of 11 motes was deployed during a testing session that lasted 96 hours.
The tested scenario included two Base Motes, eight Infrastructure Motes and two Personal Monitors (PM).
The main objective of this testbed was to test the system's reliability and effectiveness.
Screenshots and photos of the deployment is shown below.
Setup The green circles represent Infrastructure Motes while the blue circles represent roaming Personal Monitors. One Infrastructure Mote and one Personal Monitor in an office environment. Multiple Infrastructure Motes along a corridor.
Code
The developed software can be found in download section.
People Principal Investigator: Funding This work is supported in part by the NSF
grants CNS-0509268 and CNS-0237884.
Upon detecting a fall, the PersonalMonitor will send an Emergency Notification packet to the base station.
The base station can then choose to open a VoRT to allow the patient to request for help or notify
that the detected event was a false alarm.
The PM microphone is sampled at 3.33 kHz. Although the recommended sampling rate for human voice
is at least 4 kHz, the received voice is still recognizable. Currently, we are not able to sample at
higher frequency due to MICAz hardware limitation.
The implemented FallDetector follows a simple algorithm and relies on MICAz accelerometer.
The FallDetector always buffers a one second window of the wearer's orientation.
If an impact (acceleration of 1.8 G) is detected, FallDetector will wait for further impacts within 2 seconds. After 2 seconds with
no extra impacts, the FallDetector will compare the orientation of 1 second before the first impact occurred
and the current orientation. A fall is detected if the change in orientation is higher than 47 degrees.
A smarter fall detection mechanism has been proposed by utilizing Bayesian Network concepts.
This proposal is detailed
in this thesis.
ALEIS also performs localization based on RSSI and the deployed Mote Infrastructure. A Personal Monitor
broadcasts a localization beacon every 5 seconds, and each Infrastructure Mote that receives it will stamp the
RSSI and send the packet upstream to the base station. The MonitoringSystem will then track the location
of the PersonalMonitor. The implemented localization strategy simply associates the Personal Monitor to
the Infrastructure Mote with the highest RSSI.
The MonitoringSystem is a software residing at the Assisted Living Environment. It continuosly collects
data like PM location and emergency notification.
The MonitoringSystem can also be controlled remotely by agents from a healthcare provider.
Marco Caccamo (website,
email)
Students:
Chin F. Cheah
Bach Duy Bui
Andrew Tzakis
Rodolfo Pellizzoni
Copyright © 2007 University of Illinois
- Questions? email: 
Last Updated: 5.10.07