Anomaly Detection in BACnet systems
Meeting 26
Title: Clustering, generating anomalous data
Date: 31/03/2020
Time: 4 pm
Location: Zoom Conference Call
Attendees: Sofian Ghazali, Rahul Balamurugan, Zahid Kamil
Absences: None
Action Items: Finalize on generating anomalous data and clustering analysis
Meeting Agenda:
-
Create anomalous data
-
clustering analysis
Discussion and Notes:
In this meeting, the anomalous data needs to be generated to test our model. We decided to move on with what we believed to be right which is to generate a csv file that has all IP addresses different from the ones existing in the BACnet data and the occurrences of the messages similar to that of the original dataset. Therefore, we can see that the features of the binary digits can be really be experimented with and see if our binary-encoded categorical data was exactly what we did. Therefore we decided to move this type of data and then move to the testing phase. The cluster analysis was determined on how well would our clusters be able to detect any anomalies from the generated anomalous data. Therefore, we need toset a threshold on how we are going to detect these anomalies.
Next meeting action items:​
-
Determine which anomaly detection method would best fit on detecting anomalies
-
Move to testing phase
Meeting Adjournment: 6:30 pm
Meeting Minutes prepared by: Zahid Kamil