Enhancing Network Error Detection and Anomaly Identification through Visual Analytics

The increasing adoption of Building Automation Systems (BAS) or Building Management Systems (BMS) underscores the necessity for enhanced and standardized communication protocols to facilitate seamless integration and compatibility among diverse system components. As BAS networks grow, traditional methods for identifying errors and anomalies often struggle to fully grasp the intricate patterns within data, thereby hindering effective detection of anomalies or errors within data traces. This challenge is compounded by the expanding volume of network traffic data in BAS, leading to many anomalies escaping human detection. Consequently, there is a pressing need for innovative and efficient techniques to address this issue. This thesis delves into the exploration of how Visual Analytics (VA) can enhance the detection of network errors and the identification of anomalies. It primarily aims to support network experts in the analysis of networks, with a specific focus on the Building Automation and Control Networks (BACnet) protocol. Modern network analysis tools operate within three distinct areas. Firstly, automatic diagnostic tools analyze captured network traffic to detect and report network problems. Secondly, top-level analysis focuses on aggregated data sets. Lastly, in-depth analysis delves into the specifics of individual packets. While some tools excel in one area, they often lack in others. This work seeks to integrate all these capabilities into a single tool to enhance error detection across a broader spectrum. Initially, foundational aspects such as an introduction to BACnet and visualization techniques were explored. Experts were consulted to identify potential errors and anomalies and a comprehensive catalogue was compiled. From this compilation, concepts were formulated on how to visualize the data effectively. Users have the capability to import Packet Capture (PCAP) data and explore it visually through the application, uncovering new insights. Finally, a detailed error report was presented and users are provided with actionable suggestions on how to address these detected issues. The application underwent testing and evaluation by experts, revealing its capability to aid users in identifying issues and extracting valuable insights from the data.