Discipline: Computer Sciences and Information Management
Subcategory: Computer Science & Information Systems
Brandon Joyce - University of North Carolina at Greensboro
It has become increasingly complex to detect intrusion through network records due to large volumes of network traffic data. In order to detect anomalies and do it quickly, parallel strategies such as MapReduce can be used to improve the performance. In this project, network data was modeled as a graph in order to apply a graph compression algorithm. A star graph configuration model allows for the use of a pattern recognition algorithm to identify network intrusions. We have developed a MapReduce algorithm that uses a SUBDUE graph compression method. We tested our algorithm on the 1999 KDD Cup network intrusion dataset. It has shown that our MapReduce algorithm is capable of processing large amounts of data with high sensitivity (99.21%) and accuracy (94.65%). This indicates that many network anomalies can be identified using our MapReduce algorithm.
Funder Acknowledgement(s): This research is funded by National Science Foundation CCF-1460900 under the Research Experience for Undergraduates Program, awarded to Enyue Lu, Associate Professor for Computer Science and Mathematics Department, Salisbury University, Salisbury, MD.
Faculty Advisor: Enyue Lu, EALU@salisbury.edu
Role: I did the entire research project (programing, algorithm development, calculating results, etc.) with guidance from my mentor.