Anomaly Detection of MIL-STD-1553 Word Types Based on LSTM

MIL-STD-1553 is a data bus standard that is widely used in many military scenarios due to its high reliability and stability for real-time time-division multiplexing communication. The robust design of the MIL-STD-1553 helps to improve the fault tolerance of the bus system in the event of an anomaly, but detection of these anomalies is necessary to minimize their generation and ensure reliable data transmission from a testing perspective. With a variety of information transfer formats on MIL-STD-1553, it is important to find a method of anomaly detection that is accurate, efficient and universal. To address this issue, this paper proposes a method based on Long Short-Term Memory (LSTM) to predict MIL-STD-1553 word types as an indicator of anomaly detection. Firstly, this paper proposes a MIL-STD-1553 word encoding method to extract the sequential features during information transmission. Then, an LSTM-based model is used to predict the type of the next word based on the encoded sequence of known MIL-STD-1553 words. Finally, the anomalies are determined by comparing the actual word type with the predicted word type. The method was validated on datasets containing word type anomalies and datasets containing attack injections that can lead to word type anomalies. The experimental results show that the method is effective in identifying word type anomalies in MIL-STD-1553 information transfer sequences.