Research
I worked as Graduate Research Assistant in CS department. I am Interested in Computer Security in terms of exploring various ways to make a stable and secure system that does not compromise with performance. I was working with Dr. Luc Longpré and Dr. Eric Freudenthal in Robust Autonomic Systems’ Group.
We try to maintain our "secure" state stable by enforcing a defined "web of trust" where we roam. But the web in general is purely anonymous and contains potential threats for any individual. My current interest is focused on Trust Management, or various ways to ensure that a trusted entity can be trusted easily & seamlessly until the credentials proving his trustworthiness expires. Also, Trust management systems to date are designed in a tight coupling mode and I am trying to provide a more generic interface that can be coupled with any existing system. Furthermore, while the existing systems provide a transactional trust model resembling a time-of-sale type requirement, I am working to on a prolonged trusted relationship that will be maintained until the credentials expire.
My research work is described in the steps I followed,
This was my first initiative towards access control mechanism, specially Role Based Access Control or RBAC. I came across dRBAC, distributed Role Based Access Control, paper and wanted to understand the underlaying mechanism. So, I started with writing an API in java that can be used as a module in systems that want to have RBAC as access control mechanism. The result was JaRBAC, a configurable java library for access controller to enable RBAC mechanism.
Doubt:
This is the follow on from dRBAC with the hope that the credential validation can be robust using a credential dissemination system, Fern. We hoped Doubt can help distributed systems to reliably trust others by providing distributed access control mechanism using dRBAC supported by scalable, authenticated and updatable dictionary to disseminate credential informations.
VPAF:
"Doubt" would have been a near perfect solution for access control mechanisms in distributed systems, but in real life we need more flexibility. Validated and Prolonged Authentication Framework is latest initiative that promises more freedom for collaborative distributed systems. More information can be found in publications page, in the recent publication on same topic.
Recently I have been working on fault tolerant systems. Web services today, requires a lot of performance hacks to cope-up with the ever-growing usage and complexity of various web-apps. Furthermore, an emerging technique, namely cloud computing, where eventually the end-user work on a thin client, pushing all the computational complexity to the web (so called cloud), requires a lot of caching to reduce frequent query to the data server. I have been using memcached to design such caching requirements in a manner that the developer need not to worry about availability of the cache. Its still in research stage and I am busy with stack of papers :)
Also, a recent trip to NSDI'10 was really a great experience for me to have a deep insight on top things (researches) happening in these field.