Heterogeneous anomaly detection from network traffic streams using data summarization

Author Identifiers

Liam Riddell


Date of Award


Degree Type

Thesis - ECU Access Only

Degree Name

Master of Computing and Security by Research


School of Science

First Advisor

Mohiuddin Ahmed

Second Advisor

Paul Haskell-Dowland


The extreme volumes of modern networks and the increasing demands on security professionals present a critical need for analysis efficiency. Network anomaly summarization combines the broad threat detection characteristics of anomaly detection with the big data reducing qualities of summarization. However, summarising anomalies from network traffic data streams presents numerous obstacles. This thesis proposes a novel attack to anomaly mapping technique for heterogeneous network threat classification and provides a novel auto-encoding latent reflection approach for summarising network anomalies. Key findings include several new heterogeneous anomaly variants, promising performance of the novel summarization method, and the shortcomings of existing evaluation metrics.

Access Note

Access to this thesis is embargoed until 28th November 2023.

Access to this thesis is restricted. Please see the Access Note below for access details.