Revealing resilience: AI anomaly detection driven design considerations for Cyber Physical Systems supporting critical infrastructures in Small Island Developing States

Ensuring the resilience of Cyber Physical Systems (CPSs) has emerged as a critical priority, particularly for small island developing states (SIDS) and low-resource regions such as the Caribbean. These systems, which tightly couple sensing, computation, and control, are increasingly vulnerable to operational disruptions such as cyber-attacks and sensor faults. In this context, resilience must go beyond disturbance tolerance to encompass real-time anomaly detection (AD), system adaptation, and recovery, ensuring sustained operational safety, reliability, and continuity. This paper uses a machine learning based AD case study to examine the resilience of a Vehicular Cyber-Physical System (VCPS) configured for platooning. Rather than proposing a novel detection method, the study leverages an established framework to extract deeper insights into the resilience needs of VCPSs operating under constrained conditions. For regions with limited redundancy and prolonged recovery times, such as SIDS, the findings emphasise the importance of integrated detection mechanisms that identify threats and support timely and adaptive system responses. This work positions anomaly detection as a diagnostic lens for resilience, contributing to sustainable, secure, and trustworthy transportation infrastructure aligned with broader development goals for CPSs.