This proposal describes my research and education plans for the next five years and beyond. The research plan draws upon my five years of industrial experience in Bell Communications Research (Bellcore). During my employment with Bellcore, I worked on a number of projects related to communication networks, protocols, and architectures. I have worked on both the existing narrowband networks such as the Common Channel Signaling network used in the telephone network as well as the ATM-based broadband networks and cellular and wireless networks. As a part of my dissertation work and post-doctoral research, I have worked on parallel and distributed systems including scheduling in multiprocessor systems. The proposed research plan builds upon these areas of expertise to investigate some important problems in next generation high-speed networks.
The research plan has two main objectives: 1) design and analyze network control algorithms, 2) apply parallel and distributed computing methodologies to design scalable server architectures for high-speed networks. The class of network control problems that will be investigated in this research are those that arise when either due to a large number of connection request or due to faults or failures in network elements, some application level resource becomes a target of a focused overload. The outcome of this research will be algorithms for automatic detection of such focussed overloads and mechanisms to control them either by throttling the traffic close to the source or by dynamically replicating resources to meet the demand. This research effort will also result in the understanding of the interaction between the control algorithms at the different layers in the protocol stack. The goal of the second research objective will be to apply parallel and distributed computing principles in the design of scalable and fault-tolerant servers which are required for protocol processing functions and as information repositories in high-speed networks. The expected outcomes are novel caching and replication strategies and solutions of associated performance issues such as the load balancing problem. This research will also identify the trade-offs that are associated in implementing high-speed protocol processing systems using shared memory and distributed memory parallel architectures.
The education plan focuses on developing new curriculum, both at the undergraduate and the graduate levels. The new courses will emphasize experimentation and prototyping. Furthermore, these courses will develop both the fundamental principles on which the modern communication networks are built as well as the key areas of expertise that the industry needs to deploy services and applications on the next generation high-speed networks. The education plan also includes ideas for the development of courses that cater to the needs of the industry, particularly in the context of re-training and continuing education. The final part of this education plan involves active participation in a few specific outreach programs both within the university and the surrounding community. The goal of these outreach programs is to ensure that the information is available to everybody and to demonstrate the opportunities in a global information network.
This proposal, if funded, will enable me to contribute to some of the challenging problems in the next generation high-speed networks by building up on my expertise. It will also enable me to develop a strong teaching plan that is personally rewarding and at the same time achieve the goals that an academic job needs to fulfill both within the university as well as in the surrounding community.
Research
Dipak Ghosal 
Networks
Lab
This page is maintained by
Dipak Ghosal