Multi-modal Sensor Fusion for Cyber-resilient Navigation of Autonomous Ships

Commercial seagoing vessels, including container ships, bulk carriers, and tankers, play a crucial role in global transportation. However, they are vulnerable to various forms of cyberattacks and electronic interference, driven by motives ranging from industrial espionage and economic sabotage to piracy and terrorism. The consequences of these attacks can be severe, posing significant threats to the economy, human safety, and the environment. Given the shipping industry’s key role in global supply chains, and envisioning the imminent emergence of Maritime Autonomous Surface Ships (MASS), the risk of substantial economic and ecological harm from such attacks is high. Recognizing this, governments and organizations have prioritized maritime cybersecurity over the past two decades, emphasizing the need for effective protection, with most of the efforts so far being concentrated around securing the network infrastructure against a security breach. This thesis, delves into the advancements in MASS navigation systems, identifies research gaps and proposes solutions to enable shipborne navigation systems resiliency to cyber-incidents from an internal point of view, i.e., beyond and within the limits of a perimetric security breach. The research methodology achieves this by adopting an information perspective, combining practical applications with theoretical analysis, emphasizing the limitations of the established navigation instrumentation and motivates for the adoption of additional sensing modalities, such as optical sensors, to augment the cyber resilience capabilities of navigation systems. The highlighted findings of this thesis include innovative approaches that enable the detection and mitigation of Global Navigation Satellite System (GNSS) spoofing and jamming attacks. The efficacy of these methods is validated through comprehensive analysis of both real-world data collected in maritime environments and while also simulated scenarios. Maritime Autonomy is an extremely rapidly shape-shifting engineering discipline without well established methods and practices. Given the exponential traction that MASS have gained over the past years, the author considers the exposition of the research output and the accommodating engineering practices in this thesis, a noteworthy contribution to the field, offering insight to future researchers and practitioners. In Chapter 4 the technological contributions that enable the development of Situational Awareness (SA) on MASS are reviewed, from the point of view of their sensing capabilities. Besides hardware integration, developing MASS technology requires a significant investment in specialized software development knowhow, which is presented in Chapter 5. Chapter 6 presents the key engineering and organizational challenges encountered in the project’s development process, along with strategies around overcoming them. Besides technical contributions, the thesis identifies scientific reearch gaps related to the existing methodology, and addresses them by presenting the author’s research contributions in the attached publications. In Chapter 7, the thesis reviews the use of Sensor Fusion (SFU) and Multi-Modal Machine Learning (MMML), as frameworks that increase the amount of information extracted from heterogeneous data sources, enable alternative means of Positioning, Navigation, and Timing (PNT), and enhance SA. In Chapter 8, enabled by the frameworks expanded in the previous chapter, the thesis presents a novel approach to the design of cyber-resilient shipborne navigation systems, based on the consideration of shipborne navigation as a Cyber-Physical System, and the combination of Multi-Modal Sensor Fusion (MMSF) with statistical change detection, to detect and counteract cyberattacks.