At Virginia Tech I work under the supervision of Prof. Wu-chun Feng, Dr. Mark Gardner and Eric Brown. I work in collaboration with Juniper Networks to study and develop models for next-generation networks as part of my doctoral studies. For this research I was awarded a graduate research assistantship. I also work with Advanced Research Computing (ARC) as helpdesk support.
Computing devices and communication hardware have evolved dramatically over the years — hand-held devices are ubiquitous, most devices have multiple network interfaces and network infrastructure hosts numerous middleboxes to support enhanced communication. However, as network communication software has not evolved at the same pace, application developers bear the burden of managing modern use cases, which results in duplication of effort. While the networking community today is focused on research that enables programming of the network (i.e., Software Defined Networking) it is equally important to develop a data plane that allows application developers to leverage available resources and exploit them to the fullest.
The wide spread and highly successful deployment of the legacy assumptions in the Internet, makes further evolution difficult. Though substantial research has been done to extend TCP, yet most solutions have not been adopted. This is perhaps because the cost of transition from the incumbent technology to the new technology out weighs the value offered — either the proposals are not backward compatible and the transition cost is high in comparison to the value offered; or the lack of backward compatibility precludes incremental adoption altogether; or the solutions are limited in scope such that the value offered is far less than the transition cost. Therefore TCP’s success in itself has become an impediment for its evolution. We believe that is is possible to make strategic changes that re-enable the ability to innovate while retaining compatibility with the current Internet.
We are working towards a communication model that defines abstraction for the data plane, which not only enable richer constructs (e.g., sessions, flows), but also facilitates innovation in communication methods (e.g., resilient communication, hybrid transport). We plan to derive an architecture from the model that characterizes the roles of the components in the network stack and defines the interfaces between the components. We also intend to implement a prototype framework that applications developers can use to build their software.
I work with Advanced Research Computing (in the role of helpdesk support) to assist students, faculty and staff in troubleshooting problems when using the supercomputers at Virginia Tech. This involves troubleshooting problems in executing parallel code/simulations as well as packaging, testing and deploying necessary software packages. I am also responsible for diagnosing miscellaneous problems which are then forwarded to system administrators for resolution.
SLAC National Accelerator Laboratory
I was responsible for conducting research and development activities under a joint research collaboration – Measurement and Analysis of the Global Grid and Internet End-to-End Performance (MAGGIE) – for the Internet End-to-end Performance Measurement (IEPM) project along with providing technical support and coordination for the collaboration between National University of Sciences and Technology (NUST), Pakistan and SLAC National Accelerator Laboratory (SLAC), Stanford University. I was partially responsible for maintaining the Ping End-to-end Reporting (PingER) infrastructure. In this context I had setup a PingER archive site at NUST, Pakistan. In addition to that I had also setup a PingER-like infrastructure for the Pakistan Education Research Network (PERN) -- this infrastructure was aimed at providing a performance monitoring platform to coarsely measure end-to-end connectivity between the universities.
Here I was actively involved in the design and development of a decision theoretic approach to detect anomalies in Internet paths using end-to-end performance measurements. The proposed algorithm detects significant events by monitoring the available bandwidth. Typical path characteristics are extracted from bandwidth estimates. A decision theoretic approach is then applied to detect events by comparing the observed measurements to the baseline model.
I also implemented three reporting tools for the presentation of PingER measurement results (see Toolbox at the WIKI e.g., PingER Metrics Motion Charts). I am also supporting graduate and undergraduate students in their research on the Trilateration Utility for Locating IP Hosts. The aim of this project is to develop an algorithm which employs PingER and Planetlab nodes deployed worldwide to determine the geolocation of a public network interface (IP address) (see TULIP).
I have also been responsible for assisting Dr. Les Cottrell in compiling the annual reports for the International Committee for Future Accelerators – Standing Committee on Interregional Connectivity (ICFA-SCIC) Monitoring Working Group.
Kyung Hee University
I developed and implemented a solution for (application layer) issues of intermittent connectivity for mobile/hand-held devices in wireless networks. This was achieved by developing a programming library which provided seamless transition between online server components and offline stubs to the client applications in the face of varying network conditions such as fluctuating network bandwidth and disconnections.
I also designed and developed a middleware architecture that extended Grid services to hand-held devices. The middleware addressed the issues of job delegation to selected services, support for offline processing and interaction with heterogeneous mobile devices. This was primarily achieved by outsourcing the resource intensive tasks from the mobile device to the middleware using the surrogate model.
National University of Sciences and Technology
In my role as a Lecturer in the Communication Systems Engineering Department at SEECS, NUST, I was responsible for tutoring undergraduate courses on subjects such as Computer Networks, Distributed Systems and Enterprise Application Development. In addition to that I was responsible for planning and collaborating research activities for the project MAGGIE as well as providing technical support to research scholars.
Pakistan Telecommunication Company Limited
As part of my senior design project at National University of Sciences and Technology, I designed, developed and implemented a distributed network analysis system for specialized analysis of ‘Voice over Internet Protocol’ (VoIP) traffic on a data network. Specialized analysis of SIP and H.323 standard was conducted by software developed in C using tools such as KDevelop and KDE/Qt for the Linux environment. The system is currently in use for monitoring, managing and authorizing service providers. This project was developed under an agreement reached between NUST and Pakistan Telecommunication Company Limited (PTCL) and is in use today.