Embedded
Networks Laboratory
Computer
Science Department
University of
Southern California
Los
Angeles, CA 90089
Telephone: 1-213-821-5627
Fax: 1-213-740-7285
Email: 
|
|
Xin Li
Embedded
Networks Laboratory Telephone: 1-213-821-5627 |
I am a system-track Ph.D. candidate. My thesis advisor is Professor Ramesh Govindan. I am working with the group Embedded Networks Lab.
My areas of general interest include computer networks, operating systems, distributed systems, and wireless sensor networks. I have been intensively working with distributed network anomaly detection systems and distributed indexing and query processing systems.
Back to the year 2003, I started with the project DIM which aimed at building a distributed range query supporting system for sensor networks. This is the first distributed storage system that supports multi-dimensional range queries for wireless sensor networks, and has inspired many follow-up research efforts. DIM divides the data space into indexed blocks and assigns each block to a sensor node. Sensor readings are stored in the node according to the index and can be fast retrieved by user queries. I designed the indexing and query algorithms and evaluated with real-world sensor readings with C++ and ns-2. I also implemented DIM on the MICA mote platform and publicly demonstrated the prototype in several venues. This work is published in ACM Sensys 2003.
Soon after DIM, we recognized that the similar idea can also be used for resolving range queries on the Internet. Then distributed network anomaly detection was selected as the driven-application. This formed a new project, MIND. This is the first operational distributed indexing system specifically designed and deployed to support network monitoring on the Internet. It allows fast routing of multi-dimensional tuples and queries. It balances traffic and storage load in order to minimize search time and risks of congestion It is robust to attacks and network failures. I designed and evaluated the main algorithms with real Internet traffic. I also implemented a prototype of the system with Java. This prototype has run on the planet-lab with over 100 nodes across the world. This work is published in IEEE INFOCOM 2006 and IEEE NetDB 2005.
After MIND, I keep working along the direction of distributed network-wide anomaly detection and has designed schemes for this purpose.
Many outside people are involved in this long-term project including Dr. Wei Hong (Co-founder of Arch Rock Corporation), Dr. Christophe Diot (Director of Thomson's Research Lab at Paris), Dr. Gianluca Ionnaccone (Senior researcher at Intel Research Cambridge), and Dr. Hui Zhang (NEC labs America).
