Anomalous event detection for pinger data

Kayathiri, Batumalai (2015) Anomalous event detection for pinger data. [Final Year Project Report] (Unpublished)

	PDF Kayathiri(24pgs).pdf Download (6MB)
	PDF (Please get the password by email to repository@unimas.my , or call ext: 3914 / 3942 / 3933) Kayathiri(fulltext).pdf Restricted to Registered users only Download (35MB)

Abstract

The PingER (Ping End-to-end Reporting) Malaysia is an extension of the main PingER project with the objective to monitor and collect data of internet performance specifically for sites in Malaysia and also to cover the region of South East Asia. PingER Malaysia initiative is collaboration between SLAC (Stanford Linear Accelerator Center), National University of Sciences and Technology, University Malaysia Sarawak, University Technology Malaysia and University of Malaya. Apart from the main goal of data collection and monitoring of Internet performance, another equally important goal of the PingER Malaysia initiative is to create as much as research opportunity as possible in order to solve existing problems, and/or improve current performance one of which is anomalous event detections for PingER data. This project investigates into detecting anomalous event found on PingER data by calculating standard average between two points and comparing it with threshold value through step-by-step phases: (I) define research problem, (2) review the literature, (3) formulate hypothesis, (4) design research, (5) collect data, (6) analyze data and (7) interpret and report via Research methodology. At the end of this project, an anomalous event detection framework that able to display result of number of anomalous event detected on extracted PingER data monitored from University Malaysia Sarawak using pinger.unimas.my node located at latitude of 1.4653 and longitude of 110.4274. The framework is also able to visualize anomalous event detection result in table and graph through a simple web interface for reporting purpose by PingER system administrator(s}. This project is certainly important to allow human intervention onto PingER project in order to take necessary measurement by knowing and analyzing anomalous event taking place on time series graph of PingER project as each anomalous event reflect a defect during the transmission of packets from pingER. unimas.my to different sites around the world that can be a disaster to network performance.

Actions (For repository members only: login required)

View Item