Abstract
The automatic identification system (AIS) reports vessels’ static and dynamic infonnation, which are essential for maritime traffic situation awareness. However, AIS transponders can be switched off to hide suspicious activities, such as illegal fishing, or piracy. Therefore, this paper uses real world AIS data to analyze the possibility of successful detection of various anomalies in the maritime domain. We propose a multi-class artificial neural network (ANN)-based anomaly detection framework to classify intentional and non-intentional AIS on-off switching anomalies. The multi-class anomaly framework captures AIS message dropouts due to various reasons, e.g., channel effects or intentional one for carrying illegal activities. We extract position, speed, course and timing infonnation from real world AIS data, and use them to train a 2-class (nonnal and anomaly) and a 3-class (nonnal, power outage and anomaly) anomaly detection models. Our results show that the models achieve around 99.9% overall accuracy, and are able to classify a test sample in the order of microseconds.
Controlled Indexing
- learning (artificial intelligence) ,
- marine engineering ,
- neural nets ,
- pattern classification ,
- security of data ,
- ships
Non-Controlled Indexing
- machine learning-assisted anomaly detection ,
- maritime navigation ,
- automatic identification system ,
- maritime traffic situation awareness ,
- AIS transponders ,
- illegal fishing ,
- illegal activities ,
- multiclass artificial neural network ,
- multiclass anomaly framework ,
- AIS message dropouts
Introduction
Maritime security is of utmost importance today as over 90% of international trade as well as thousands of passengers around the world are carried over sea [1]. Maritime transport poses significant challenges, natural as well as humaninduced, e.g., tough and unpredictable environment, collision, illegal fishing, smuggling, pollution, and piracy. Nowadays, the automatic identification system (AIS) has become an essential part of maritime traffic situation awareness. Vessels equipped with AIS transponders report their positions, which are based on the global navigation satellite system (GNSS), their navigational status, as well as other voyage related information. This information can be used in collision-avoidance mechanism, tracking of vessels, detecting unusual trajectories of vessels, etc. However, AIS transponders can be switched off to hide suspicious activities, such as illegal fishing, or piracy. Therefore, it is essential to use real world AIS data in order to analyze the possibility of successful detection of various anomalies in the maritime domain.
Anomaly detection in maritime traffic has attracted researchers to apply various statistical and machine learning solutions [1], [2]. Statistical solutions, such as extended Kalman filter and particle filter, have been used to reconstruct trajectory of vessels [3]. When the estimated and real trajectories diff er more than a predefined threshold, those events are categorized as an anomalous behavior. Bayesian network was applied for tackling missing values (anomalies) in a dataset for prediction and classification purposes [4]. A hidden Markov model was utilized to detect AIS on-off switching (OOS) anomaly with he consideration of transmission channel characteristics [5]. Machine learning (ML) algorithms, such as artificial neural network (ANN), support vector machine (SVM), long shortterm memory (LSTM), on the other hand, have been shown to perform better than statistical methods for prediction and classification problems in general. Recently, [6] proposed a one-class SVM-based anomaly detection framework that takes AIS data as well as received signal strength into consideration for analyzing AIS OOS anomaly. [7] used a random forest ML algorithm to classify vessels using AIS data streams. A long-short-term-memory (LSTM) algorithm was used in [8] to reconstruct vessels’ trajectories. Although the use of ML models for anomaly detection has just started recently, ANN, which can learn complex fitting functions and has been shown promising as compared to other ML techniques, has not been applied to deal with the multi-class intentional and nonintentional AIS OOS anomaly yet. A multi-class model is vital to capture AIS message dropouts due to various reasons, for example, channel effects (e.g., power outage) or intentional (for illegal activities). We expect more AIS dropouts, when vessels move away from the AIS receiver. Thus, it is important to distinguish between the intentional and natural AIS OOS. The natural dropouts of AIS messages can be identified using the vessel’s distance from an AIS receiver where the weak received signal strength may result in loss of messages.
In this paper, we propose an ANN-based anomaly detection framework to detect an AIS OOS anomaly. We use realworld AIS messages to train our multi-class anomaly detection framework, and test on further real AIS data. More in detail, we extract a four-dimensional (4-D) feature vector containing latitude, longitude, speed, and course information from each AIS message transmitted by vessels. After resampling the received AIS messages, we train the ANN models with data samples containing one or more 4-D feature vectors within an observation period. A data sample is labeled as an anomaly when it has consecutive AIS dropouts more than a predefined threshold. First we train and test a 2-class model consisting of normal and anomaly samples. Our results show that the 2-class AIS OOS model can detect anomaly correctly with an overall accuracy of nearly 100%. In order to distinguish intentional AIS OOS anomaly from the AIS dropouts due to weak received signal strength, we add “power outage” as another class. For this purpose we add the distance as an extra feature, calculated between the position of vessels and a receiver station, to classify power outage samples based on a predefined AIS transmission range. Upon training the 3-class anomaly model with this additional feature, the model achieves around 99.9% overall accuracy, and that too in a reasonable time of a few microseconds per sample. The rest of this paper is organized as follows. Section II briefly describes the AIS and its usefulness in the anomaly detection framework. Section III presents a neural network model for anomaly detection and data preparation steps for the models. We evaluate the performance of the models in Section IV, and conclude the paper in Section V.
Conclusion
In this paper, we proposed a multi-class artificial neural network (ANN)-based anomaly detection framework to classify intentional and non-intentional AIS on-off switching anomaly. The multi-class anomaly model captures AIS message dropouts due to channel effects and intentional one. We utilized position, speed and course, and timing information from real-world AIS data for the training of the multi-class anomaly detection models. Our results show that the models achieve around 99.9% overall accuracy, and misclassify only a few samples. The anomaly detection framework is designed in such a way that we can further classify the anomaly types, for instance, an anomaly when vessels are moored, make U-turns, or enter into a restricted zone. For future work, we would apply our models to live maritime traffic scenarios.
About KSRA
The Kavian Scientific Research Association (KSRA) is a non-profit research organization to provide research / educational services in December 2013. The members of the community had formed a virtual group on the Viber social network. The core of the Kavian Scientific Association was formed with these members as founders. These individuals, led by Professor Siavosh Kaviani, decided to launch a scientific / research association with an emphasis on education.
KSRA research association, as a non-profit research firm, is committed to providing research services in the field of knowledge. The main beneficiaries of this association are public or private knowledge-based companies, students, researchers, researchers, professors, universities, and industrial and semi-industrial centers around the world.
Our main services Based on Education for all Spectrum people in the world. We want to make an integration between researches and educations. We believe education is the main right of Human beings. So our services should be concentrated on inclusive education.
The KSRA team partners with local under-served communities around the world to improve the access to and quality of knowledge based on education, amplify and augment learning programs where they exist, and create new opportunities for e-learning where traditional education systems are lacking or non-existent.
FULL Paper PDF file:
Machine Learning-Assisted Anomaly Detection in Maritime Navigation Using AIS DataBibliography
author
Year
2020
Title
Machine Learning-Assisted Anomaly Detection in Maritime Navigation Using AIS Data
Publish in
2020 IEEE/ION Position, Location and Navigation Symposium (PLANS), Portland, OR, USA, 2020, pp. 832-838,
Doi
10.1109/PLANS46316.2020.9109806.
PDF reference and original file: Click here
Somayeh Nosrati was born in 1982 in Tehran. She holds a Master's degree in artificial intelligence from Khatam University of Tehran.
Professor Siavosh Kaviani was born in 1961 in Tehran. He had a professorship. He holds a Ph.D. in Software Engineering from the QL University of Software Development Methodology and an honorary Ph.D. from the University of Chelsea.
Nasim Gazerani was born in 1983 in Arak. She holds a Master's degree in Software Engineering from UM University of Malaysia.
