Analysis of RF Signal Anomaly Detection Performance Using Leaky Integrate-and-Fire Neuron-Based Spiking-NN in Space Propagation Environments

This study evaluates the RF link anomaly detection under space propagation variations. Noise, signal loss, and interference were progressively added to the European Space Agency (ESA) satellite telemetry data, and the performance of a Leaky Integrate-and-Fire (LIF)-based spiking neural network (SNN) was compared with those of the long short-term memory (LSTM) and 1D-convolutional neural network (CNN) models. While the performance differences were minimal under normal conditions, the SNN showed the smallest F1-score degradation (2.6 %) as environmental variations increased, outperforming LSTM (10.1 %) and 1D-CNN (44.2 %). These results highlight the importance of robustness, and indicate that LIF-based SNN is effective for bit error rate (BER) management and operational availability in satellite communication systems.