LATEST (INTEGRATED) PROJECT WEB PAGE

Publications: Recent papers | ATM Forum Contributions | Old Papers and Technical Reports

• Rate-Based Flow Control Algorithms for High Speed Networks: An NSF funded project with Prof. Hitay Özbay to study control theoretic underpinnings of rate-based congestion control algorithms and to develop rate-based schemes for the Internet. DARPA sponsored a related piece of work on bit-based edge-to-edge congestion control.

  This work is an extension of my Ph.D. dissertation work at Ohio State University (OSU) was in the area of traffic management in ATM networks, where my advisor was Prof. Raj Jain. We continue to work closely with OSU group and the ATM Forum, the IETF , and several industries, besides publishing our work in journals, workshops and conferences.

  We see applicability of rate-based schemes in edge-to-edge (intra-domain) control of aggregated flows at the core of the Internet where architectures such as multi-protocol label switching (MPLS) . We also use rate-based control to modify TCP windows and otherwise control TCP performance, in an approach similar to that of Packeteer Inc. . We are also applying rate-based control to multicast congestion control in another project. We work closely with the IETF ( diff-serv and MPLS groups) and several interested industry partners. Related Work .

• A Comparative Performance Study of TCP Rate Control: A Packeteer sponsored project to study the performance of TCP rate control versus alternatives such as Random Early Detection and Explicit Congestion Notification. Related Work.

• A Study of TCP/IP and ATM over ADSL : A Pulsecom sponsored project involving a study of schemes for improving TCP/IP performance over the Asymmetric Digital Subscriber Line (ADSL) technology. Related Work.

• Study of the differentiated services architecture for the Internet: Awaiting Nortel and NSF support for the current project which studies TCP-friendly building blocks (traffic conditioning, provisioning, edge-to-edge flow control) of the differentiated services (diff-serv) architecture for the Internet. We work closely with the IETF and several companies in this effort. This project also has synergy with the Internet pricing project where we look at pricing as another mechanism for traffic management. Related work.

  Our long-term goal is to study and develop the QoS-related components needed to build the next generation of scalable virtual private networks (VPNs).