Dependability Prediction with Logical Link Correlation for Industrial Wireless Communication Systems

Dependability is critical for industrial wireless communication systems because of the high requirements of industrial applications. As an important phase, dependability prediction can foresee future states of a wireless communication system and can support control and improve the state of the wireless communication system. Recent dependability prediction methods predict the states only based on time series which ignores the logical link correlation. Therefore, this paper proposes a sequence to sequence model based on long short-term memory and an attention mechanism to leverage the logical link correlation to improve prediction accuracy. We conducted comparative experiments on realistic measurement data sets with three cases, where the proposed model always outperforms the benchmark, which proves the novelty.