Measurement level AIS/radar fusion

Abstract

In maritime surveillance, messages from radar and the Automatic Identification System (AIS) receivers are used for vessel trafficking and monitoring. The common trend is to use radars as the primary source of surveillance and AIS as a secondary source with little interaction between these data sets. The AIS messages provide very accurate position estimates associated with ID and other vessel information. However, AIS messages arrive unpredictably and intermittently depending on the type and behavior of the vessel. In addition, the revisit interval of AIS messages could be very large and it may vary from one vessel to another.In this work, a new measurement-level fusion algorithm to combine radar and AIS messages is proposed using the Joint Probabilistic Data Association (JPDA) framework. The proposed method handles AIS ID swaps between vessels and missing IDs while effectively fusing the radar measurements with AIS messages at measurement level. The uncertainty in the AIS ID-to-track assignment is resolved by assigning multiple AIS IDs to a target and updating the ID probabilities using a Bayesian inference with radar measurements, AIS messages and other targets. The performance of the proposed measurement-level fusion is compared with that of the track-to-track fusion. A modified Posterior Cramér-Rao Lower Bound (PCRLB) is also derived for the variable-rate heterogenous AIS/Radar network. Experimental results based on simulated data demonstrate the performance of the proposed technique.