A framework for damage detection in AIS data based on clustering and multi-label classification

Automatic Identification System (AIS) is a telecommunication system that allows marine equipment (ships, shore -based stations) to communicate and transmit information regarding their trajectory (position, speed, course, etc.), so that other ships are aware of their presence and collision between vessels can be avoided. Unfortunately, AIS struggles against some technical limitations (relatively short range of its terrestrial segment that led to the introduction of satellite AIS segment, that deals with new problems, namely packet collisions) that leads to parts of the AIS data being damaged during transmission. Therefore, to maintain the proper level of maritime safety and security, there is a need for reconstruction of such damaged AIS data, so that ships keep receiving appropriate updates from their neighbours. In this paper we propose a machinelearning based approach for AIS data reconstruction, with one of its stages - detection of damaged data that requires reconstruction - being its main focus. The characteristics of the proposed approach is presented, together with the results of several computational experiments that measure its performance. This article is an extended version of a paper that was presented at the International Conference on Computational Science in 2023 (ICCS-2023) (Szarmach and Czarnowski, 2023), enriched with the results of additional computational experiments.