Outlier Detection in Wireless Sensor Networks
Outliers in data mining and statistics is a well known and well studied problem. Knowing the outliers in one's data set is quite useful since it can detect a myriad of events ranging from security compromises to emergency situations to failing hardware. Classical techniques in outlier detection don't work in wireless sensor networks. The unique challenges of communication, limited resources, and the geographic distribution of data sets in wireless sensor networks turn this classical problem into a hard problem. I am currently developing a new technique that will detect outliers with all these considerations in mind.
Researcher
Dylan McDonald
Advisor
Dr. Sanjay Madria
|