-
Protecting Location Privacy using Hard-to-Reverse Negative Databases
As GPS sensors become cheaper and more widely available, several states governments have proposed or are currently using these devices to monitor people that have
commited various crimes. In this way, parole officers can ensure that parolees stay within known bounds. Also, the location of people on parole can be correlated with
locations where crimes have been recently committed. Unfortunately, not very much attention has been given to the privacy rights of the parolees. We propose
using hard-to-reverse negative databases to protect the privacy of parolees. Users will be able to query the negative database for specific locations at specific
times, but unable to query for an exhaustive list of locations. In this way, parolees can still be correlated with known crime events, while maintaining their
privacy.
Collaborators: Eric Trias, Elena Ackley, and Stephanie Forrest
-
A Filesystem Interface for Sensor Networks
Sensor network users currently face enormous challenges, including programming difficulty and the use of unfamiliar, complex interfaces.
To date, most usability research with respect to sensor networks have focused on simplifying programming by offering powerful programming
abstractions. Unfortunately, such work does little to encourage ordinary users, such as application scientists, to adopt sensor networks. In order
to address these issues we have developed a filesystem interface for sensor networks. By treating a sensor network as a standard Unix filesystem,
users are able to use familiar tools to interface with sensor networks. This simplifies management and debugging. Similarly, programs
originally designed for filesystems, such as file-sharing programs, can be used to extend sensor networks. Finally, users are also able to
prototype applications using pre-existing programming environments that interface with the sensor network through the file I/O interface.
Collaborators: Jean-Charles Tournier, Patrick Widener, and Ann Kilzer
-
Tables: A Table-based Language Environment for Sensor Networks
Intuitive programming environments that target application specialists and casual users are currently lacking. Current tools either requires large investment
from users to learn advanced programming techniques or focuses on simplistic and limited tools. Tables is a graphical programming environment that addresses this problem by
utilizing a spreadsheet-inspired programming interface that interacts with a local runtime executing on sensor nodes. Tables emphasizes ease-of-use by reusing spreadsheet abstractions, such
as pivot tables and functions, to interactively program the sensor network. By using these familiar tools, users are able to construct complex applications that include local data
filtering and collective processing.
Collaborators: Eric Nelson
Negative Survey: Anonymous Data Collection in Sensor Networks
Sensor networks involving human participants will require privacy protection before wide deployment is feasible. Providing this privacy has typically involved the use of crytographic approaches
that must maintain complicated infrastructures and protocols for key maintainence. The negative survey is an alternative method for privacy protection that can be used in certain applications with a much lower
computational, memory, and infrastructure overhead. Sensor nodes, instead of transmitting their actual data, transmit a sample of the data complement to a basestation. The basestation then
uses the negative samples to reconstruct a histogram of the original sensor readings.
Collaborators: Michael Groat, Stephanie Forrest, and Fernando Esponda
Kensho: A Dynamic Tasking and Deployment Framework for Sensor Networks
Kensho is a library designed to distribute a set of programs onto a sensor network (tasking) efficiently using the underlying
mechanisms provided by the operating systems. The library offers a set of functions to task the network using a variety of methods,
including spatial and temporal methods. Unlike distributed programming abstractions, Kensho actively attempts to separate
the role of tasking a sensor network from programming a sensor network application. This allows sensor networks
equipped with Kensho to use multiple programming abstractions simulataneously.
Collaborators: Angela Mielke
Kaizen: Improving Sensor Network Operating Systems
Kaizen is a software framework for physical resources management on sensor nodes. The
purpose of Kaizen is to provide and guarantee access to physical resources in order to
maintain a given level of quality of service to each executing task. Kaizen can be seen
a thin software layer between applications and the kernel, and provides a set of interfaces
to safely use the physical resources formalized using contracts.
Collaborators: Jean-Charles Tournier