Newsletters:
Summer 2003 | Spring 2004 | Fall 2004 | Summer 2005
There is an every growing interest and need to deploy sensors
for monitoring our various infrastructure systems for the
purpose of condition monitoring, operational control, and
security. However, in these infrastructure applications,
we simply will never be able to replace sensor systems every
three years the way we do for today's existing computer
systems. And while no one single component is expected to last
100 years, it is the overall system that must remain functional
and provide the necessary information over that timeframe. The
overall system must be able to function over the 100-year
lifetime of the system, and must not be dependant on keeping a
large stockpile of spare parts or antiquated applications
available.
The Institue for Complex Engineered Systems (ICES) at Carnegie
Mellon University has established a major
interdisciplinary research program for agile evolutionary
infrastructure sensor networks. This program capitalizes on the
CMU tradition of multidisplinary research and education in order
to create a framework for applications of networks of sensors in
long-lived infrastructure systems such as buildings, bridges,
and highways, as well as power, telecommunications, and
transportation systems. Infrastructure systems are not
completely replaced, but are replaced in a piecemeal
manner. Remodeling occurs in a building room-by-room and a
section at a time for a highway system. When a sensor system is
integrated with the infrastructure, it is desirable to replace a
piece of the sensor system at that time. The net effect is a
heterogeneous collection of sensors that must continue to
operate.
The SensorNets project has four main thrusts:
Devices, Applications, Systems, and Data.
Devices
Critters: Pervasive Sensor Networks
Critters is the name of the project as well as the name of
the scalable pervasive sensor networks device developed at CMU.
The Critters project is building CMU as The Most Sensed
Campus by pervasively deploying thousands of sensors around
campus. The initial version was implemented by the students of
39-606
Spring 2003, which include Yen-Chun Cho, Shakul Tandon, and
Ahmed Wahedi. This group was led by Mike Bigrigg. Since then we
have expanded upon the critter to include an automated
calibration device (patent pending) and a light-version of the
device. As part of the SUREthing summer
program, Nick Dasaia worked on a microphone-port version of the
Critter device, and Steven Frank worked on the USB port.
Commercialization of the Critter is being pursued by Pervasive Sensors.
Please contact them for commercial product support and sales.
Room Sensor
The Room Sensor is being pursued by the students of 39-606
Spring 2004, which include Hung-Fu Chang, Mike Duke, Sarah
Knepp, and Vipul Singvhi. The group is supervised by Mike
Bigrigg. This project is sponsored by Microsoft Corporation in
the Embedded Devices Group.
Rack Sensor
Currently being investigated by Mike Bigrigg, CMU undergrad Greg
Jaworski, and CMU INI student George Nychis.
Wireless Energy Meter
A project supervised by H. Scott Matthews.
Micro-Fuelcell Wireless Sensor
Initially developed by SUREthing student Steven
Frank. Additional development by CMU undergrad Greg Jaworski.
Supervised by Mike Bigrigg and S.C. Yao.
Systems
cAEIN, Agile Evolutionary Infrstructure sensor Networks
The cAEIN project recognizes the need for an integrated approach
to long-lived systems. Components include adaptable
evolutionary software development, multi-dimensional modeling
and analysis, and sensor
data quality assessment.
FlakyIO, Reliable I/O Systems
As I/O devices (disks, network interface cards, handhelds,
sensors) gain processing power, they also gain an ability to use
that processing power for reliability self-awareness. These
devices can report exceptions to respond appropriately to the
internal I/O failures. We use exception
injection to explore the ability of systems to appropriately
respond to exceptions raised by these self-aware devices. This
project has been worked on by many people.
Vizit, Millions of Data Samples Visualization Explorer
The understanding of data is not a simple matter of drawing a
graph, but allowing the engineer to interact with the data to
understand patterns and trends. In addition, the amount of data
that can be collected from sensor networks (tens to hundreds of
gigabytes) pushes the limits of most visualization systems.
ETC graduate student Mike Duke is building an interactive data
visualization system. HERE is a sample
screen shot.
Cache Optimizations for Data Intensive
Applications
There is a great disparity between improvement of CPU
performance and memory performance. RAM speeds have not been
increasing as quickly as the CPU. We took on this problem of
streaming data optimization by doing analysis of cache
utilization. We have been able to achieve up to a 50%
improvement!
Wireless Building Models
Developed initially by SUREthing student, Sean
Salem. Additional development by CMU undergraduate student, Joe
Cahill. This is a joint project between CMU, Mike Bigrigg, and Lehigh
University, Liang Cheng.
Data
One Sensor
One machine in our facilities runs a commercial temperature
sensor with the data available over the the web. You can view
its current
value or a few weeks of data
from January 2004.
The Most Sensed Laboratory
Outfitting the ERIS Reliable Systems Computer Laboratory with a
host of sensors allowed people to understand how they can
interact with pervasively deployed sensors throughout
campus. You can view their current readings HERE.
June 2003
The first collection of raw sensor data occured over one week in
the summer. It consisted of two temperature detectors placed in
each of the ICES graduate student offices and laboratories
(1208, 1210, 1212, 1214, and 1216). It recorded temperature
readings 10 times a second. This data is not processed or
cleaned. It consists of the actual reading as reported by the
detector. Many commercial products will employ algorithms to
clean the data. We left it in to allow our colleagues who do
data mining and other processing to work with the raw
data. RawData1 is 2GB of readings and is available by sending
email to Mike Bigrigg (bigrigg@cmu.edu). A technical report
which describes the data is available as a PDF file CMU-04-01.
West Wing
The Institute for Complex Engineered Systems occupies the first
and second floor of the west wing of Hamburg Hall at CMU. The
Heinz School occupies the basement floor. We are distributing
sensors through the floors of the West Wing to gather a larger
data set to explore the correlation of data in different yet
similar locations. This data set is starting its second year of
collection. Look for its release soon!
Applications
39-605 : Spring 2003 : Decentralized Machine Room Monitoring
Important computer servers are no longer being isolated in
climate controlled machine rooms. Instead, many servers are
being distributed through an organization yet have the same
requirements for an ideal environment, such as temperature.
There is a need for decentralized monitoring yet centralized
control of the rooms that house important machines.
39-650 : Fall 2003 : Building Management
The building facilities management staff at CMU receives 50-60
complaints a day mostly about temperature issues. There is a
tremendous overhead associated with sending a person to a
location to determine if there is an HVAC (heating,
ventillation, air conditioning) problem. The goal of this
project is remote building management.
39-650 : Fall 2003 : Energy Monitoring
In order to implement an effective energy management program to
encourage energy conservation, it is important to measure and
understand the energy consumption (lighting, heating and
cooling). This project should help to identify wasteful energy
practices as well as a means to quantify improvements.
39-650 : Fall 2004 : Security
Security has become an increasingly important issue. The
pervasive use of sensors (door/window, occupancy, cameras, etc.) has
sparked the need for better access and management of these systems.
Projects included security for the home, office, and also
infrastructure.
39-650 : Fall 2004 : Transportation Monitoring
39-650 : Fall 2004 : Disaster Management and Investigation
The students in these projects used sensors on mobile robots as well
as wireless sensors to be used to help EMS during a fire and for
investigators after a fire.
IUP COSC320 : Fall 2004 : Building Management
The Indiana University of Pennsylvania will be using the CMU
Critter System as part of their Software Engineering Practicum
course. The topic is building management systems.
IUP COSC320 : Spring 2005 : Energy Monitoring
IUP will again be using the CMU Critter System as part of their
Software Engineering Practicum course. The topic area this time
is energy monitoring systems.
The SensorNets research programs reach across the Embedded and Reliable
Information Systems (ERIS) and Advanced Infrastructure
Systems (AIS) laboratories within the Institute for Complex Engineered
Systems at CMU.
Support for many of the SensorNets research projects is provided
by the Pennyslvania Infrastructure Technology Alliance (PITA).
Additional support provided by Microsoft Corporation. Software has
been graciously donated by Veritas
Corporation.
Please direct any questions or comments to Mike Bigrigg, Project
Scientist, 2210 Hamburg Hall, Institute for Complex Engineered
Systems, Carnegie Mellon University, Pittsburgh PA 15213. You
can also reach him at 412.268.7668 or bigrigg@cmu.edu.