Princeton University

School of Engineering & Applied Science

Aikaterini Varsou
Graduate Student
Graduation Year: 
Thesis Title: 
Rate Scheduling Techniques and Quality of Service Guarantees for Heterogeneous Networks
Vincent Poor
Current Employer: 
IPVALUE Management Inc.

09/05 till now] Member of Technical Staff, IPValue Management, Inc. Working on commercializing intellectual property for Xerox partner. Responsible for Web and Ecommerce program.
[03/01 - 09/05] Research Staff Member, NEC Laboratories America, Inc., Broadband and Mobile Networking Group (formerly IP&Optical Networking Group), Princeton, NJ.
[08/05 –09/05] Presence Information Presence information and its subset of Instant Messaging are an emerging technology area. Presence denotes the availability and willingness of a user (presentity) for communication. The user’s information is published to certain subscribers (watchers). The main protocol proposed is SIMPLE, an extension to SIP, while there is also open source XML-based protocol, maintained by Jabber. My work initially consisted of a literature search on presence applications and papers as well as a market analysis of the different products available, comparison of the several protocols and their features. Then certain architectural problems were addressed in an effort to favor distributed or centralized systems. Context-awareness was introduced and researched in an effort to simplify things. My main task is to address privacy filtering in a presence context through use of specially modified in-house technology.
[04/05 –09/05] RFID Radio Frequency Identification, its application, standardization, and innovation are constantly changing. Its adoption is still relatively new and hence there are many features of the technology that are not well understood. Developments in RFID technology continue to yield larger memory capacities, wider reading ranges, and faster processing. It is believed that it is highly unlikely that the technology will ultimately replace the bar codes, since even with the inevitable reduction in raw materials coupled with economies of scale, the integrated circuit in an RF tag will never be as cost-effective as a bar code label. However, RFID will continue to grow in its established niches where bar code or other optical technologies aren't effective. If some standards commonality is achieved, whereby RFID equipment from different manufacturers can be used interchangeably, the market will very likely grow exponentially. In the new fiscal year I started researching RFID. After conducting a survey on the literature and a market and competitor analysis, I identified specific problems to work on based on the major “bottleneck” categories: security, collisions, blocking, yoking proofs, as well as using stream processing and complex event processing to attack RFID issues. This is a brand new project and ongoing research.
[2004] Meta Data Syndication for Grid/Web The Meta Data Syndication (MDS) project is a multi-year project in NEC Labs to research on fundamental issues pertaining to data syndication and distribution, using the proliferating messaging paradigm of publish and subscribe (pubsub) systems. These areas include, among others, distributed routing algorithms, high performance content matching/filtering/processing, reliable messaging, ordered delivery, robustness and scalability, security, business models with regards to accounting, etc. The MDS project aims to provide a unified system for large amount internet scale applications such as RSS feed distribution, Network Monitoring, etc, which can be extended to support monitoring, surveillance and control applications, resource brokering, edge-based admission control, etc. MDS is based on an asynchronous messaging based pubsub system that allows communication between entities through exchange of messages while knowing each other in advance. This property allows building a Service Oriented Architecture (SOA) for many Internet-scale applications. It also allows optimization in the areas of routing, caching, network load balancing, content matching/filtering, security, etc. for such applications. MDS will investigate topic and content based pubsub systems and their applications. MDS chooses XML as its wire format and supports SOAP overlay or native TCP/IP overlay as its networking primitives. I got involved with this new and continuing project and had several accomplishments: from identifying potential RSS/MDS issues, to designing a portal and a website for the syndication project to finally simulating in ns2 our proposed solutions. Several technical reports were written and publications are currently being prepared.
[2003 till now] Grid Networking (content distribution for broadband and wireless) The Grid networking philosophy targets viewing and utilizing network, computing and storage resources as a single virtualized system by discovering, monitoring and opportunistically harvesting available resources. Grid networking is being developed as a mechanism for several types of applications: data storage and replication, software distribution, rich data or content distribution and managed data delivery. As part of the Grid Networking team, I investigated existing schemes and solutions and extended the project by providing innovative algorithms for the provisioned case. I built a simulation platform to model realistic networks and measure the performance and optimization of the new algorithms based on multiple metrics and wrote up supporting documentation, manuals and technical reports presenting the results. I conducted a competitors comparison, investigated existing patents and technologies as well as market forecasts. I advanced the Grid project by proposing an application to storage networks, which I pursued through analysis and simulations. Relevant publications are currently being prepared for submission. This is a patent pending project, on which I worked in 2003 and which still continues as more relevant issues arise and as it gets integrated with the Web services and MDS topics. For details, please check the publications section.
[2002] Hierarchical Optical Networks My work focused on routing algorithms and issues for the NEC Labs HOXi project. I extended the project by considering "hot" topics such as protection for the optical networks and creation of an equivalent virtual topology. I also developed a software tool simulating a HOXi network to visualize the algorithmic performance. I created a parser that interacted with the core code and provided all the supporting documentation and reports. For details on this project, please check the relevant publications in the appropriate section.
[2001] High Speed Routers My work focused on new packet forwarding techniques and their comparison to the existing ones. I performed an exhaustive literature search and gave an internal seminar on different approaches for the problem.