Indexing Multi-Dimensional Data in Peer-to-Peer Systems

Project: Research project

Project Details

Description

IIS-0534343 Wang-Chien Lee Pennsylvania State University, University Park Indexing Multi-Dimensional Data in Peer-to-Peer Systems The goal of this project is to provide support for efficient search of multi-dimensional data in Peer-to-Peer (P2P) systems. The project achieves its goal using the following approaches: (1) Develop a new P2P overlay, called semantic small world (SSW), to serve as a general infrastructure (which includes a network, a distributed index, as well as a logical file structure) for supporting a variety of queries at the same time; (2) Develop an innovative dimensionality reduction method that maps data clustered in high-dimensional space to a low-dimensional SSW to reduce the high maintenance overhead required for storing high-dimensional data in constantly changing P2P environments; (3) Develop search algorithms for point query, range query, K nearest neighbor (KNN) search, Top-K query, similarity search and joins based on SSW; (4) Prototype SSW and the search algorithms on PlanetLab; and (5) Develop simulators and analytical models to evaluate the performance of the developed techniques. The research results will have great impacts on distributed data management in e-commerce, information retrieval, digital government and scientific studies. This project supports Ph.D students to pursue research in the areas of distributed data management and P2P systems. A new graduate-level course covering these areas and integrating the research results from the project will be introduced into the curriculum. Publications, technical reports, software and experimental data from this research will be disseminated via the project web site (http://www.cse.psu.edu/pda/P2P).

StatusFinished
Effective start/end date1/15/0612/31/09

Funding

  • National Science Foundation: $350,000.00
  • National Science Foundation: $350,000.00

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.