The ship trajectories collected by the Automatic Identification System (AIS) are widely used in maritime applications. However, a significant issue with AIS data is that large AIS gaps occur. Existing trajectory imputation methods for AIS data have three main limitations: (1) the temporal aspect is ignored; (2) the methods fall short when dealing with complex ship movements; (3) the common-route assumption does not always hold. To overcome these limitations, we propose TrajImpMC, a tracking-based framework that uses polygon-based ship location estimates from multiple cameras to impute large AIS gaps. TrajImpMC combines speed constraints and Kalman filters, and can return imputed trajectories that contain both spatial and temporal information. Extensive experiments are conducted on real datasets. In terms of the quality of the imputed trajectories, TrajImpMC improves the RMSE errors by at least one order of magnitude over two existing state-of-the-art AIS imputation methods. In addition, a visual comparison shows that the imputed trajectories of TrajImpMC align very well with the real ship trajectories during AIS gaps. The code for this paper is available at: https://github.com/songwu0001/TrajImpMC.

The wide spread of Automatic Identification System (AIS) has motivated several maritime analytics operations. Vessel Location Forecasting (VLF) is one of the most critical operations for maritime awareness. However, accurate VLF is a challenging problem due to the complexity and dynamic nature of maritime traffic conditions. Furthermore, as privacy concerns and restrictions have grown, training data has become increasingly fragmented, resulting in dispersed databases of several isolated data silos among different organizations, which in turn decreases the quality of learning models. In this paper, we propose an efficient VLF solution based on LSTM neural networks, in two variants, namely Nautilus and FedNautilus for the centralized and the federated learning approach, respectively. We also demonstrate the superiority of the centralized approach with respect to current state of the art and discuss the advantages and disadvantages of the federated against the centralized approach.

Recent advances, especially in deep learning, allow to effectively detect ship targets in surveillance videos. However, the translation of these detections to the real-world locations of ships has not been sufficiently explored. The common approach in the literature is using a transformation matrix to convert a pixel to a real-world coordinate. However, this approach has three shortcomings: first, a set of reference point pairs has to be manually prepared to establish the matrix; second, the matrix always maps a pixel to the same real-world coordinate, ignoring that there is no one-to-one correspondence between discrete pixel coordinates and continuous real-world coordinates; third, this approach can only work with one camera. In light of this, we propose a technique PixelToRegion that explicitly takes into account the uncertainty in coordinate conversion by mapping each pixel to a spatial polygon. Next, we propose a new algorithm MCbSLE that can estimate ship locations using pixel sets from multiple cameras. The precision of location estimation by MCbSLE is enhanced through spatial intersection between polygons from different cameras. Experiments are conducted under 16 carefully designed multi-camera settings to evaluate MCbSLE w.r.t. four factors: different ports, the number of cameras, the distance between cameras, and camera headings. Results on one-day ship trajectory data show that (1) an 79.8% accuracy in the number of coordinates can be achieved by MCbSLE when there are no more than 10 ships in camera views; (2) using multiple cameras can improve the precision of location estimation by one order of magnitude compared with using one camera.

The rapid development of autonomous transportation systems is currently motivating several research initiatives. This paper presents an approach for efficient task allocation and path planning for an Unmanned Surface Vehicle (USV) which is requested to transfer products in an island group in the Aegean Sea. The proposed approach is based on a bi-level scheduling methodology, in the upper level, considering the weather conditions and the geographical characteristics of the area we create a square (from/to) travel time matrix. In the lower level, considering travel times we develop a Genetic Algorithm to achieve the USV’s task allocation and path planning. Experimental results demonstrate the effectiveness of our approach in guiding the USV’s to efficiently accomplish short sea container transshipment in the island group of Cyclades.

Digital transformation strategy (DTS) involves redesigning various organisational operations to encompass digital technologies and achieve business objectives. In this study, we explored digital strategy and value creation shifts in omnichannel organisations that aimed to deliver seamless online and on-premises (dual-mode) customer experiences. Using comprehensive data on the gambling industry, we focused on the long-term effects of dynamic relationships among multiple DTS events over time. Building on digital strategy and value-creation theory, we observed and analysed organisational changes linked to technological shifts in omnichannel organisations during turbulent times and disruptions. Herein, we discuss the balance between online and on-premises service channels in terms of a DTS pathway, viewing it as a dual-mode value-creation process. By exploring this dual-mode value-creation process, we contribute to DTS theory and omnichannel operations. Furthermore, we enhance theory by unveiling the impact of shifting digital strategy perspectives on the transformation of omnichannel organisations in dynamic and disruptive business environments. We also present strategic propositions for planning and realising DTS requirements for omnichannel service providers in a broad context.

Monitoring vessel traffic on a global scale is a complex and challenging task. The large number of moving vessels and the complexity of monitoring their position and forecasting their route in real-time require novel, advanced and highly scalable big-data mechanisms. In this work a digital twin for constant maritime situational awareness on a global scale is presented. The described multi-layered system is able to visualize maritime traffic in real-time, based on data from the Automatic Identification System (AIS), while also providing forecasts of future movement based on machine learning and deep learning techniques. The system is validated using real streaming AIS data from around the globe to demonstrate its performance, scalability and parallelization efficiency.

We present a Maritime Situational Awareness (MSA) framework for detecting and forecasting maritime events (e.g., illegal fishing) over streams of Big maritime Data. The architecture of the MSA framework relies on the following state-of-the-art components: (i) the Maritime Event Detector which uses data-driven distributed techniques deployed on a computer cluster to detect maritime events of interest in an online, real-time fashion, (ii) the Complex Event Forecasting module, which implements state-of-the-art distributed Complex Event Forecasting techniques for maritime data, (iii) the Synopses Data Engine component, that creates synopses of maritime data improving the scalability of the framework and (iv) the streaming extension of a popular data science platform, namely RapidMiner Studio, that integrates all the above, allowing users to graphically design and rapidly implement Big Data analytics pipelines which can be deployed transparently on top of distributed architectures.

Nowadays, the increasing number of moving objects tracking sensors, results in the continuous flow of high-frequency and high-volume data streams. This phenomenon can especially be observed in the maritime domain since most of the vessels worldwide are now transmitting their positions periodically. Therefore, there is a strong necessity to extract meaningful information and identify mobility patterns from such tracking data in an automated fashion, eliminating the need for experts' input. To this end, a novel approach is presented in this paper, which fuses the research fields of computer vision and trajectory classification, in order to deliver a high-precision classification of mobility patterns. The experimental results demonstrate that the classification performance of the proposed approach can reach an f1-score of over 95%.

The article “MongoDB Vs PostgreSQL: A comparative study on performance aspects”, written by Antonios Makris, Konstantinos Tserpes, Giannis Spiliopoulos, Dimitrios Zissis, Dimosthenis Anagnostopoulos, was originally published electronically on the publisher’s internet portal on 05 June 2020 without open access.

Abstract Several modern day problems need to deal with large amounts of spatio-temporal data. As such, in order to meet the application requirements, more and more systems are adapting to the specificities of those data. The most prominent case is perhaps the data storage systems, that have developed a large number of functionalities to efficiently support spatio-temporal data operations. This work is motivated by the question of which of those data storage systems is better suited to address the needs of industrial applications. In particular, the work conducted, set to identify the most efficient data store system in terms of response times, comparing two of the most representative of the two categories (NoSQL and relational), i.e. MongoDB and PostgreSQL. The evaluation is based upon real, business scenarios and their subsequent queries as well as their underlying infrastructures and concludes in confirming the superiority of PostgreSQL in almost all cases with the exception of the polygon intersection queries. Furthermore, the average response time is radically reduced with the use of indexes, especially in the case of MongoDB.

Complex Event Processing (CEP) 's main purpose is recognizing interesting phenomena upon streams of data. So its only natural that it would find applications in the maritime domain, where detecting vessel activity plays an important role in monitoring movement at sea. In this study we briefly examine the field of Complex Event Processing; we present two CEP implementations, one based on machine learning techniques and a rule-based system modeled with Event Calculus. Finally, we evaluate their ability in modeling activities that involve multiple vessels, by comparing their results on real-life examples.

In this paper we attempt to define the major trade routes which vessels of trade follow when travelling across the globe in a scalable, data-driven unsupervised way. For this, we exploit a large volume of historical AIS data, so as to estimate the location and connections of the major trade routes, with minimal reliance on other sources of information. We address the challenges posed due to the volume of data by leveraging distributed computing techniques and present a novel MapReduce based algorithmic approach, capable of handling skewed and nonuniform geospatial data. In the direction, we calculate and compare the performance (execution time and compression ratio) and accuracy of several mature clustering algorithms and present preliminary results.

Seaports play a vital role in the global economy, as they operate as the connection corridors to all other modes of transport and as engines of growth for the wider region. But ports today are faced with numerous unique challenges and for them to remain competitive, significant investments are required. In support of greater transparency in policy making, decisions regarding investment need to be supported by data-driven intelligence. It is often an overlooked fact that seaports do not remain static over time; such spatial units often evolve according to environmental patterns both in size but also connectivity and operational capacity. As such any valid decision making regarding port investment and policy making, essentially needs to take into account port evolution over time and space. In this work, we leverage the huge amounts of vessel data that are progressively becoming available through the Automatic Identification System (AIS) and distributed machine learning to define a seaport's extended area of operation. Specifically, we present our adaptation of the well-known KDE algorithm to the map-reduce paradigm, and report results on the port of Shanghai.