MaDMAN : Middleware for Delay-Tolerant Mobile Ad-Hoc Networks
The MaDMAN (Middleware for Delay-Tolerant Mobile Ad-Hoc Networks) project focuses on delay-tolerant networks (DTNs), or environments of mobile, intermittently connected nodes. We are working on an adaptive middleware which will allow applications in these environments to automatically adapt their communications to suite their changing environment.
The following architecture diagram illustrates our general concept. In it, the Device has access to multiple independent "network stacks", possibly utilizing different transport, network, link, and physical protocols. The Middleware multiplexes application-layer connections across these stacks in response to changing network conditions, and migrates active connections between stacks as the situation demands.
For more information about the MaDMAN middleware and links to download our code releases, please see the MaDMAN Project Website
I am also involved with the Pharos Mobile Computing Testbed, a modular testbed here at the University of Texas comprised of mobile, autonomous robots (such as the one pictured below). Pharos provides a platform for running real-world pervasive computing experiments on heterogeneous mobile devices.
Passive Contxt Sensing for Context-Aware Mobile Computing
As computing devices and their users become increasingly mobile, the demand for information about the application's environment, or context, becomes significantly important to the efficient and robust operation of mobile and pervasive computing systems. Applications must be able to adapt themselves to changing conditions to satisfy users' demands and expectations and to ensure that the application's resource usage matches the environment's capabilities. Sensing context using traditional means incurs network communication, which competes with the applications using the network and expends valuable network resources, especially communication bandwidth and battery power. This project explores passively sensing context metrics. This results in measurements that are basically approximations of actual context, but can be collected with zero cost in terms of network communication. This project develops a model of passive context sensing and a general framework for building and deploying passively sensed context metrics.
The Passive Context Sensing Suite
We have built a collection of passive context sensing modules implemented in C++ for the Click Modular Route . It is available to download: Click PCS Suite.
Please note that the code is a prototype, and as such it might not be stable under all platforms or experimental setups.
During my undergraduate studies, I took part in the UT Honeynet Project (now defunct).
We helped develop and test some of the early open-source contributions of the Honeynet Project, an international security research organization which gathers data about malicious hackers, their tools, methodologies, and motivations.