Anomaly Detection for Cellular Networks using Deep Learning

This project developed a deep-learning-based anomaly detection method for the automatic detection of anomalies of cellular network KPIs.

A technical flowchart of an outlier detection pipeline. It shows an original time series ($Z$) passing through data pre-processing ($X$), TCN-AE based feature learning (encoder and decoder) to produce an output time series ($Y$), and finally threshold-based outlier detection involving reconstruction error calculation to identify outliers.

A line graph titled "Synthetic Time Series with Outliers" showing a blue oscillating signal against a time index from 0 to 700. Specific data points that deviate significantly from the pattern are highlighted with red dots to indicate detected outliers.

The diagnosis of cellular performance degradation mostly relies on manual observation and analysis of huge amount of network KPIs, which is labour-intensive and error-prone.

Admissions AY2026

Apply Now!

Anomaly Detection for Cellular Networks using Deep Learning​

This project developed a deep-learning-based anomaly detection method for the automatic detection of anomalies of cellular network KPIs.

Anomaly Detection for Cellular Networks using Deep Learning