My main research interest is to develop end-to-end network diagnosis techniques to uncover
dynamic network properties (e.g. congestion information, bottleneck equivalence, loss rates,
topology, etc.) in a near real-time fashion. This information can be used to characterize the Internet
dynamics and to dynamically adapt the control strategies of massively accessed Web servers (e.g.
congestion control, redirection strategies, etc.) for a more efficient utilization of network resources.
 
 

Boston University                                                                                                          Boston, MA
Research Fellow Fall 1998 to present                                                    Advisor: Prof. Azer Bestavros

Designed, analyzed and implemented a framework (MINT) that is capable of diagnosing Internet
dynamics. The MINT framework has been used as part of the MASS and WING networking groups
at Boston University to analyze and characterize Internet performance and to design content
delivery protocols for massively accessed web servers. This work resulted in papers at
INFOCOM'02, PAM'02, ICNP'00 and a couple of other papers under review.

Clemson University                                                                                                      Clemson, SC
 Research Assistant Spring 1996 to Fall 1997                                               Advisor: Prof. Roy Pargas

 I was responsible for implementing a Balanced Inventory Flow Replenishment System (BIFRS) for
 the Department of Defense. BIFRS is an enterprise-wide constraint management solution to optimize
 items distribution on time and at lowest total cost.

 Alexandria University                                                                                           Alexandria, Egypt
 Research Assistant Fall 1992 to Summer 1994                                         Advisor: Prof. Amin Shoukry

 Experimented with several supervised and unsupervised Neural Network models for the speaker
 independent recognition of speech utterances. Different Neural Network objective functions,
 methods of features extraction and ways of end-pointing speech utterances have been
 investigated. A new hierarchy consisting of modules of Time-Delay Neural Networks (TDNN) and
 Single Layer Perceptrons (SLP) has been proposed, analyzed and implemented. This work resulted
 in a paper at NSRC'95.