An In-memory Database for Prototyping Anomaly Detection Algorithms at Gigabit Speeds
MetadataShow full item record
The growing speeds of computer networks are pushing the ability of anomaly detection algorithms and related systems to their limit. This thesis discusses the design of the Object Database, ODB, an analysis framework for evaluating anomaly detection algorithms in real time at gigabit or better speeds. To accomplish this, the document also discusses the construction a new dataset with known anomalies for verification purposes. Lastly, demonstrating the efficacy of the system required the implementation of an existing algorithm on the evaluation system and the demonstration that while the system is suitable for the evaluation of anomaly detection algorithms, this particular anomaly detection algorithm was deemed not appropriate for use at the packet-data level.