A Novel BeiDou Satellite Transmission Framework With Missing Package Imputation Applied to Smart Ships

Abstract

Obtaining voyage data through a vessel monitoring system (VMS) and transmitting it to a remote monitoring center through satellites is an important link to realize the information perception of smart ships. However, improving data quality to ensure the integrity of information is still an open question. Taking into account the complex and harsh marine environment, a BeiDou satellite transmission link will inevitably suffer the challenge of packet loss. Therefore, this paper proposes a novel BeiDou satellite transmission framework with missing package imputation to improve the information perception ability of smart ships. In particular, we propose a two-stage missing packet imputation strategy (T-MPI). First, a bidirectional recurrent neural network (Bi-RNN) based on temporal view imputation is designed to learn the time correlation of the data. Second, considering the attributive view imputation of multivariate data, an autoencoder with a linear aggregation module (LA-AE) is proposed to capture sudden changes in the ship’s complex sailing state. An experimental platform was established to verify the developed framework, and real freighter transportation data were used to prove the effectiveness of our proposed imputation strategy under complex sailing conditions.