IBM Israel Research Seminars

Monitoring data streams in a distributed system is the focus of much research in recent years. A distributed monitoring task consists of accurately detecting, at each point in time, whether the data complies with a certain global criteria. An example of a distributed monitoring task is using agents installed on a set of routers to detect when traffic to a certain IP address raises above a predetermined threshold. Another example of a distributed monitoring task is detecting when the average temperature reading taken by sensors in a sensor network exceeds a predetermined threshold.

Most of the proposed monitoring schemes deal with monitoring simple aggregated values. More involved challenges, such as the important task of distributed feature selection (e.g., for collectively identifying spam email or for collaborative intrusion detection), or monitoring the variance in the temperature (pressure, pollution, etc) readings taken by sensors in a sensor network, still require very high communication (and power) overhead using naive, centralized algorithms.

About the Speaker
Tsachi Sharfman Graduated from the Technion's Computer Science Department (Summa Cum Laude). He has worked for nine years as a Development Team Leader and System's Architect for leading companies in Israel's Hi-Tech industry, including Vocaltec and Checkpoint. Currently he is studying at the Technion for a Ph.D degree in Computer Science. His research topic is Distributed Data Mining algorithms.