Document Type : Original Article
High Institute for Computers and Information Technology AL-Shorouk Academy
To effectively detect and identify the anomaly data in massive satellite telemetry data
sets, the novel detection and identification method based on the Auto-regressive integrated
moving average (ARIMA), Prophet, Long Short Term Memory (LSTM), and Auto-encoder
algorithms were proposed in this paper. The proposed model is used to find anomalous events
by comparing the actual observed values with the predicted intervals of telemetry data.
First, preprocessing for the raw telemetry data were Handled for the missing values using
linear interpolation. Second, Down-casting to reduce the memory storage. Based on this
symbolization, the pseudo-period of the data was extracted. Third, the Data Transformation and
Scaling to normalize the data within a particular range to helps in speeding up the calculations
were applied. Finally, the experimental results for the Prophet model show predictions with high
efficiency, stable when detecting anomalies, and requires little computational time. The results of
Prophet compared with other applied algorithms, demonstrate the effectiveness and superiority
of the proposed model.