Anomaly Detection in CDR-Based Trajectories of the Mobile Cellular Network
Name
Tiit Vaino
Abstract
Mobile data is an excellent resource to approximate people’s location and movement. This could be used by governments and private companies. For example, emergency services use people’s phone locations to know where to send help. The approximate location is derived from combining mobile and cellular network connections information with cellular network cell (mobile mast antenna coverage area) locations. Multiple location events with consecutive timestamps, when put together, make one trajectory. It is crucial to ensure that datasets are clean to prevent costly decisions based on flawed analyses.
The main issue is that cell location in a database and in real life does not match because of human error or unsynchronized databases. This thesis proposes a model for anomaly detection in CDR-based trajectories using a Trajectory Anomaly Detection with Mixed Feature sequence (TAD-FM) approach. The training and testing were done using real data with integrated virtual anomalies, where some of the cell’s locations are deliberately modified. Additionally, improvement has been introduced to the model to reduce the time complexity for training and prediction. The proposed model was capable of labelling and detecting 66%
of the cells with wrong location data as an outlier.
The main issue is that cell location in a database and in real life does not match because of human error or unsynchronized databases. This thesis proposes a model for anomaly detection in CDR-based trajectories using a Trajectory Anomaly Detection with Mixed Feature sequence (TAD-FM) approach. The training and testing were done using real data with integrated virtual anomalies, where some of the cell’s locations are deliberately modified. Additionally, improvement has been introduced to the model to reduce the time complexity for training and prediction. The proposed model was capable of labelling and detecting 66%
of the cells with wrong location data as an outlier.
Graduation Thesis language
English
Graduation Thesis type
Master - Computer Science
Supervisor(s)
Amnir Hadachi
Defence year
2024