Home
People
Research
Publications
Contact Us
|
|
INTRODUCTION
|
|
The cost of health care for our nation's aging population
has become a national concern. According to the U.S. Census Bureau, the number
of people over the age of 65 is expected to hit 70 million by 2030, having
doubled since 2000. Health care expenditures in the United States are projected
to rise to 15.9% of the GDP ($2.6 trillion) by 2010.
Recent advances in wireless sensor networks have made it possible to deploy
wearable sensors on the bodies of patients in a residential setting, allowing
continuous monitoring of physiological signals (such as ECG, blood
oxygen levels) and other health related information (such as physical activity
levels). The application of wireless sensor networks in a medical care
environment provides a unique opportunity to shift health care outside
a traditional clinical setting to a patient/home-centered setting, and to reduce
healthcare expenses through more efficient use of clinical resources and earlier
detection of medical conditions.
There remains a significant gap between the availability of the sensing
technology and our ability to bring it into general use for home medical
sensing. A medical sensing system must provide reliable and privacy-preserving
information transmission between patients' homes and the care giver.
CareNet is an integrated wireless sensor
environment for remote healthcare that addresses these critical needs.
This work was supported in part by TRUST (The Team for
Research in Ubiquitous Secure Technology), which receives support from the
National Science Foundation (NSF award number CCF-0424422) and the following
organizations: AFOSR (\#FA9550-06-1-0244), Cisco, British Telecom, ESCHER, HP,
IBM, iCAST, Intel, Microsoft, ORNL, Pirelli, Qualcomm, Sun, Symantec, Telecom
Italia and United Technologies. |
|
SYSTEM ARCHITECTURE
|
|
CareNet is built upon a heterogeneous networking infrastructure which
involves the patient data collection, transmission, and access phases, as
shown in Figure 1. Here we focus our discussion on the
networking and system design of the data collection phase.
|
| Figure 1: System Architecture |
As illustrated in the figure, a two-tier wireless network is used to
provide data sensing, collection, transmission and processing functions.
At the lower tier, a body sensor network consisting of lightweight
wearable sensors provides data sensing and transmission functions.
These sensors communicate with each other and the base-station
sensors (which are attached to the backbone wireless network)
directly using IEEE 802.15.4 wireless standard. We use Telos motes
as the hardware devices. For movement sensing and fall detection,
these motes are equipped with accelerometers and gyroscopes as shown
in Figure 2. At this tier, sensor devices are
lightweight, wearable and mobile, which also means they have low
computation, communication power and small amount of memory. So in
our design, only necessary computational and communication tasks are
implemented at these devices.
At the upper tier of the network is a multi-hop IEEE 802.11-based
wireless network which provides a high-performance backbone
structure for packet routing. We use Stargate single board
computers as the hardware devices. A picture of the backbone
router is shown in Figure 2. The backbone routers
are connected to the base-station motes which communicate with the
mobile wearable sensors directly. The Stargate board can also be
connected with a web camera and serves as a video sensor. Equipped
with IEEE 802.11 wireless adaptors, the backbone routers communicate
with each other and relay the movement sensing data as well as video
streams to the home healthcare gateway. Using IEEE 802.11
wireless communication standard, this stationary backbone structure
provides a high-performance and high-reliability packet routing
service. Since IEEE 802.11 has a larger communication range than
IEEE 802.15.4, our design also scales much better in terms of local
area communication coverage. Finally, the home healthcare
gateway serves as an interface between the patient's home and the
care giver's medical system, which processes all the sensing data
and transmits them to the remote medical care system.
|
| Figure 2: Sensor and Router Devices |
|
|
SOFTWARE DESIGN
|
|
CareNet is also built upon a multi-layered software infrastructure
based on the features and functions at each of the network tiers. The
overall software architecture are shown in Figure 3.
|
| Figure 3: Software Architecture |
We use TinyOS operating system and NesC programming language to implement
the movement data sensing at the wearable sensors and the data transmission
between the mobile wearable sensors and the base-stations. The major
functions of the wearable sensors are to sample, synchronize, and transmit
the movement data. Beacon messages are used for hand-offs between the
base-stations and the mobile sensors when the mobile sensors are moving
into or out of the communication range of the base-station sensors. A
customer-designed and -implemented protocol based on TinySec is used for
mobile sensor authentication and secure communication between mobile
sensors and the base-stations.
We use Embedded Linux operating system and the ACE programming
environment to implement the network communication among the backbone
network routers and between the backbone routers and the home healthcare
gateway. ACE is open-source software based on C++, which encapsulates OS
concurrency and network programming APIs. We take advantage of ACE's strong
communication and concurrency capabilities in our implementation. There are
two major functions implemented at this layer: as backbone routing
structure, software components are built to route and forward the video and
sensor data to the home healthcare gateway; as video sensors, video data
sampling and compression functions are also implemented. There are two
major design considerations in the backbone network routing infrastructure:
Application-level routing. We implement the routing
protocol among the backbone routers at the application level using TCP
socket. The reason behind this design choice is that it could easily deal
with the data loss and data replication in the wireless transmission. We
implement a multiple-hop packet forwarding mechanism through TCP streams in
ACE. Since the backbone network is stationary at most of the time, we use a
semi-static routing protocol which forwards packets based on a routing
table that can be either preconfigured manually or updated by HELLO
messages every hour. The routing table points out the IP address and port
number of the next hop node by searching the destination node inside the
packet header. A backbone router may need to forward the data streams from
more than one video or movement sensors; the data streams will be forwarded
simultaneously through different threads in the system. Using TCP, the
secure communication among backbone routers is implemented based on SSL.
Mobile sensor hand-off. To ensure reliable packet delivery
during the mobile sensor and base-station hand-offs, packets from the
mobile sensors will be received by all base-stations within their
transmission ranges. This means that the same data may be forwarded through
more than one base-stations (as well as backbone routers). To remove the duplicated data
packets from the backbone network, each sensor data packet is marked
with a timestamp in its packet header. Duplicated data packets that
arrive later at the queue of a router will be dropped. The remaining
duplicated and out-of-order packets will be dropped and sorted at
the home healthcare gateway.
We use the Linux operating system and the ACE programming environment to
develop the application software for the home healthcare gateway, whose
function is to collect, aggregate, analyze, and forward the video and
movement sensor data.
|
|
Experiment and Results
|
|
|
|
People
|
|
Faculty
Collaborators
- Laura Brown, Vanderbilt Homecare Services
-
Janie Parmley, Vanderbilt Homecare Services
Students
- Shanshan Jiang, PhD student, Vanderbilt University
- Yann Cao, PhD student, Vanderbilt University
-
Sameer Iyengar, PhD student, University of California, Berkeley
-
Philip Kuryloski, PhD student, Cornell University
Project Alumni
|
|
Software Release |
|
|
|
Publications and Presentations
|
|
|
|
Video Demonstration
|
|
|
|
|
|
Related Projects and Links
|
|
|
Contact:
Prof. Yuan Xue
Vanderbilt University
Phone: (615) 322-2926
Last updated October 16, 2007
|
CareNet: