Techniques and Applications of Overlay Networks

In this sub-project, we address the research issues in the scalable overlay network infrastructure and its application to continuous data stream processing.

Scalable overlay network infrastructure

In this part of the research, we will focus on the design and development of scalable overlay networks for supporting data streaming applications. We will research into related theories and algorithms, and build the infrastructure. It will contain the following major components: a middleware or run-time environment for creating and executing overlays, a development framework for developing application-specific overlays, and a monitoring and management system for the overlays.

Continuous data stream management

Based on the scalable overlay network infrastructure, we will continue to build up the basic components and for managing continuous data streams on the overlay networks. With our research experience in the fields of data broadcasting, distributed query processing and optimization, web mining, content-based multimedia retrieval, multimedia networking, parallel processing, and algorithm design, we will research into the issues on querying, mining, and delivering continuous data streams and develop efficient methods for the variety of streaming applications. Our works toward this goal are divided into two parts. In the 1st and 2nd years, we will consider the issues of continuous query indexing, time-varying summary maintenance, trend analysis, and video stream delivery. We will design the mechanisms for processing continuous/streaming queries, the one-pass mining algorithms for finding frequent patterns and data/query clusters, and the scheme for delivering video streams. In the 3rd and 4th years, we will further address the issues of the blocking operators in continuous queries, data heterogeneity, approximate matching and summarization for multimedia streams. We will construct the mechanisms for processing complex queries with data heterogeneity and the real-time algorithms for semantic object identification, high-level concept derivation and maintenance for multimedia streams.

‘@

‘@