Department: Computer Science & Engineering
Faculty Advisor(s): Tajana S. Rosing

Primary Student
Name: Akanksha Maurya
Email: amaurya@ucsd.edu
Phone: 858-261-8110
Grad Year: 2017

In an Internet of Things (IoT) system a large number of objects communicates with each other or with the user through sensors, which generates a massive amount of data within a small interval of time. A huge number of IoT applications require the user to gain more insight of the data and its characteristics. One of the very efficient methods to analyze data is through clustering. The data generated from sensors are unbounded, continuously arriving and evolving over time at high speed, which put a lot of challenges on analyzing the data due to memory and time constraints in an IoT application. Unlike clustering static data point, clustering time series data has many challenges. It is impossible to store a tremendous amount of data in the main memory. As we cannot store the data in main memory, multiple scans of data is not feasible. Therefore, for an IoT applications we need an efficient time series data clustering algorithm in terms of memory and processing requirements. The algorithm should provide timely results by performing fast and incremental processing of data points, should be able to capture the dynamic nature of the data and should also be able to detect the presence of outliers and act accordingly. In the paper, we discuss efficient time series clustering algorithms and its characteristics such as runtime and memory usage for an IoT application.

Industry Application Area(s)
Energy/Clean technology | Internet, Networking, Systems

« Back to Posters or Search Results

Contact:   researchexpo@soe.ucsd.edu   (858) 534-6068