Anomaly Detection in Mixed Telemetry Data Using a Sparse Representation and Dictionary Learning

•Anomaly Detection using DICTionary (ADDICT) is a novel algorithm for spacecraft telemetry monitoring.•ADDICT is a multivariate anomaly detection method based on a sparse decomposition and dictionary learning.•ADDICT handles mixed continuous and discrete data.•A comparison of performances of different anomaly detection approaches has been performed on a representative dataset.