A Survey of AI-Related Cyber Security Risks and Countermeasures in Mobility-as-a-Service

TL;DR

This article presents the first comprehensive review on the coupling between AI-driven MaaS design and the diverse cybersecurity challenges related to cyberattacks and countermeasures and focuses on how current and emerging AI-facilitated privacy risks and adversarial AI attacks may impact the MaaS ecosystem.

Abstract

Mobility-as-a-Service (MaaS) integrates different transport modalities and can support more personalisation of travellers' journey planning based on their individual preferences, behaviours and wishes. To fully achieve the potential of MaaS, a range of AI (including machine learning and data mining) algorithms are needed to learn personal requirements and needs, to optimise journey planning of each traveller and all travellers as a whole, to help transport service operators and relevant governmental bodies to operate and plan their services, and to detect and prevent cyber attacks from various threat actors including dishonest and malicious travellers and transport operators. The increasing use of different AI and data processing algorithms in both centralised and distributed settings opens the MaaS ecosystem up to diverse cyber and privacy attacks at both the AI algorithm level and the connectivity surfaces. In this paper, we present the first comprehensive review on the coupling between AI-driven MaaS design and the diverse cyber security challenges related to cyber attacks and countermeasures. In particular, we focus on how current and emerging AI-facilitated privacy risks (profiling, inference, and third-party threats) and adversarial AI attacks (evasion, extraction, and gamification) may impact the MaaS ecosystem. These risks often combine novel attacks (e.g., inverse learning) with traditional attack vectors (e.g., man-in-the-middle attacks), exacerbating the risks for the wider participation actors and the emergence of new business models.

Figures (5)

• Figure 1: In the MaaS ecosystem, cyber security risks manifest through diverse threats originating from adversarial entities, both within and outside the MaaS framework. These risks extend beyond conventional cyber assets to encompass socio-physical assets, such as individuals, organisations, and the data and AI assets under their purview. The complexity of this system underscores the multifaceted nature of data and AI risk generation and propagation. Decisions made by both human actors and automated systems, spanning individuals and organisations, play pivotal roles in shaping the dynamics of cyber security risks. Moreover, these risks transcend organisational boundaries and sectoral domains, permeating throughout the interconnected landscape of the MaaS ecosystem.
• Figure 2: A MaaS planner determines multi-modal journey using cost-combining, label-constrained, and multi-objective optimisation approaches based on the cyber assets of multiple service providers.
• Figure 3: Data attack vectors mainly consist of two categories, A. data privacy attacks and B. data-computer pipeline attacks. Data privacy risks include 1) profiling and inference and 2) third-party unauthorised access and data oversharing. Data-computer pipeline attacks include DoS (Denial of Service) attacks and attacks from socio-technical perspectives.
• Figure 4: High-level conceptualisation of a MaaS-style data ecosystem Cottrill-C2020
• Figure 5: Attack vectors against a machine learning MaaS controller: a) evasion data attacks that attempt to cause mis-performance in the machine learning algorithm, b) MitM attacks, b1) inferring personal information by eavesdropping on hyper-parameter or raw data exchanges, b2) model extraction attacks try to infer the overall MaaS controller model, and d) gamification attacks that use a sequence of false data to erode system wide performance.