Abstract

  Members

  Achievement

  Meeting

  Main Project

  Chinese Version

  

In this subproject, we will investigate the characteristics of an adaptive overlay network and develop some applications about continuous data stream processing on it.

Adaptive Overlay Networks

In this part, we will design and implement an adaptable overlay network to support data streaming applications. We will study related theories/algorithms and construct a framework infrastructure.

The framework will contain the following components:

  1. A middleware which can connect various devices to the overlay network
  2. An overlay structure which supports specific applications
  3. A monitoring and management system for the overlay network

Management of continous data stream

Based on the above adaptive overlay network, we will further construct a data management model to efficiently process continuous data streams. With our extensive experiences on related research domains such as data broadcasting, distributed query processing and optimization, network data mining, multimedia content retrieval, multimedia network, parallel processing, and algorithm design, we will proceed to study the advanced topics about the query, mining, and delivery of continuous data streams. We will also integrate these essential mechanisms to construct a high performance overlay network to satisfy distinct demands by various data streaming applications.

Our work plan is divided into two parts: in the first two years, we will study the problems on continuous query, automatic index maintenance, trend analysis, video streaming, and etc. We will design a query processing interface to support continuous query, a mining algorithm to immediately identify frequent items and classify them to groups, and a streaming video approach to efficiently deliver videos over the overlay network. In the latter two years, we will further analyze two problems known as computation complexity and data heterogeneity in continuous queries. We will also study the problems on similarity check and automatic summarization of multimedia streams. We will design an efficient query processing algorithm to handle complex queries with heterogeneous data. We will also propose a real-time analyzing mechanism to support advanced multimedia stream processing, such as object identification, semantic analysis, and high-level semantic-based prediction and maintenance.