Indian Ocean region faces many traditional and non-traditional security and safety challenges with the rising of illegal activities. Among them, illegal immigration, IUU (Illegal, Unreported and Unregulated) fishing, drugs and narcotic trafficking, human trafficking, and weapons smuggling can be identified as the most critical issues. Being an island nation, Sri Lanka has to ensure safety at sea by continuously monitoring activities at sea to mitigate threats. The International Maritime Organization introduced the Automatic Identification System (AIS) as a mandatory standard to mitigate maritime threats by monitoring vessels. Apart from AIS, several other monitoring systems such as Radar, Synthetic-Aperture Radar (SAR), and Satellite systems are also used to monitor the vessels. However, monitoring a wide range in the ocean with a large number of maritime activities is a challenging task without the assistance of computer-based systems and developing technologies. At present, Sri Lanka has a string of Navigational Radar base stations covering the coast of the country to monitor maritime vessels. One of the major drawbacks of such a setup is the inability to monitor and predict the movement of vessels and take preemptive actions when needed while demanding a high level of human involvement for the process to function properly. Further Sri Lankan Navy does not possess notable Surveillance Radar Systems at present, which would yield a larger range with the ability to perform object tracking, which however is very expensive. Therefore, developing an automatic maritime surveillance system can provide great assistance to continuous monitoring and detecting maritime threats.

Numerous illegal vessel behaviors which are different from each other can be observed in the maritime domain. To identify these complex vessel behaviors, researchers direct to use novel techniques such as deep learning with big data and artificial intelligence which is the current trend to detect abnormal behaviors. The type of illegal vessel behavior that was supposed to be captured by the existing models is rather elementary (eg. Risky speeds, unexpected stops, u-turns, tracks with unusual shapes and illegal zone entries). Further, most existing studies are based on maritime traffic violation detection and use only AIS to monitor vessels. However, some maritime vessels, such as small fishing boats, do not have AIS facility, and significant threats can also be observed with those vessels.  With these issues, detectable illegal behaviors of the maritime vessels are bounded in previous research. Therefore, this research aims to identify any illegal behavior of maritime vessels using AIS and Radar data to increase the safety and security of the maritime sector.  Summary of the proposed research is shown in Figure 1. A machine learning algorithm is proposed to use in this research by considering the previous studies. Features and a labeled data set will be collected with the help of maritime experts’ experience. Learning algorithm will be trained with these previous vessel data to classify normal and abnormal vessel behaviors. Finally, the trained algorithm can be used to detected illegal vessel behaviors quickly by ensuring the safety at maritime domain