Memory-Constrained Aggregate Computation Over Data Streams

Computing multiple aggregation queries over a data stream has applications in many domains: IP network monitoring, stock trading, analysis of Web logs, fraud detection in telecom networks and retail transactions, querying sensor node readings, etc. Salient characteristics of these applications include: very high data arrival rates that make it impractical to perform multiple passes over the data, hundreds of aggregation queries, and limited CPU and memory resources. As an example, consider an IP network monitoring system, which collects IP flow records exported by network routers and performs a variety of monitoring tasks like estimating traffic demands between IP endpoints, computing the top hosts in terms of IP traffic, profiling application traffic, and detecting network attacks and intrusions.

Provided by: Bell Labs Topic: Security Date Added: Nov 2010 Format: PDF

Find By Topic