
The Automatic Identification System (AIS), established as a standard by the International Maritime Organization (IMO) in 2008, mandates that commercial vessels over 300 gross tonnage carry AIS transceivers. These devices regularly broadcast dynamic voyage data (such as position, speed, and course) along with static vessel information (such as type and dimensions). Originally developed as a safety measure to prevent maritime collisions, AIS has since become a cornerstone of vessel monitoring and global maritime intelligence platforms like www.marinetraffic.com.
However, AIS has limitations due to its cooperative nature, as it depends on the vessel’s crew to actively transmit data. This makes it susceptible to disruptions, including equipment failure, data collisions in congested areas, or intentional deactivation. Vessels may deliberately go “dark” to hide illegal activities such as smuggling, unauthorized transshipment, or entry into sanctioned regions. As a result, relying solely on AIS leaves critical gaps in maritime situational awareness, particularly in high-risk or remote areas.
To address these limitations, the MT Tracker project proposes a hybrid tracking approach that combines AIS with non-cooperative sensing. Specifically, it leverages the fact that vessels keep their radar systems on for navigational safety, even when AIS is off. The project involves developing a robust edge computing device capable of capturing and processing both AIS and radio frequency (RF) signals emitted by vessels. By fusing these two data sources in real time, the system can detect and track vessels even when they attempt to avoid detection through AIS suppression. This approach enhances the coverage, reliability, and security of maritime monitoring, supporting efforts in surveillance, compliance, and risk management.
Technologies used

Declarative Querying

Online Data Aggregator

Edge-driven Federated Learning

Visual Analytics
What has MobiSpaces improved?
Performance in re-scheduling the service to peak period has been improved in the following Key Performance Indicators:
>94 %
Data to object association accuracy
>93 %
Data to track association accuracy
>93 %
Data to server flow reduction