I. Introduction
Cyber-Physical Systems (CPSs) are essential enablers for Industry 4.0, where physical processes are integrated with virtual equivalents for increased flexibility, performance, and reliability [1]. In order to allow seamless connection and operation of devices and services, the Industrial Internet of Things (IIoT), Artificial Intelligence (AI), and several computing paradigms (i.e., Edge, Fog, Cloud) are used within CPSs. Due to their complexity and heterogeneity, CPSs can be vulnerable to a variety of cyber and physical threats of both random and malicious origin. Exposure to those threats might have serious consequences, ranging from financial loss to loss of human lives [2]. Therefore, in order to reduce the risk of malfunctions, both the cyber and physical parts of those systems must be carefully monitored [3], possibly employing techniques based on intelligent fault detection and predictive analytics, such as predictive maintenance (PdM), which are in the main focus of this paper.