The time spoofing detection method based on Long Short-Term Memory network

Temporal and spatial information security is fundamental to the safety of national critical infrastructure. Disruption or interference with the time system can cause significant economic losses to the nation, and even pose a substantial threat to defense security. Existing timing deception detection methods primarily establish models based on the characteristics of changes in the receiver’s clock model to detect deception. However, due to the uncertainty of attack methods and the system errors inherent in the established receiver clock model calculation and fitting process, accurate fitting of the clock model parameters is difficult, and the environmental adaptability is low. To address this, this paper proposes a timing deception detection method based on the Long Short-Term Memory (LSTM) network. This method does not require consideration of the attack methods of timing deception, and has strong generalization capabilities. By utilizing the excellent time series prediction ability of LSTM, the method accurately tracks the trend of changes in receiver clock differences before and after timing deception based on the characteristics of these changes, achieving effective detection of timing deception interference. Finally, experiments and analyses are conducted using TEXBAT (Texas spoofing test battery) timing deception scenario data, and a comparison is made between LSTM and Multilayer Perceptron (MLP) networks. The results indicate that the performance of LSTM timing deception detection is superior to that of MLP.