In Cyber-Physical Systems (CPSs), devices interact with smart applications by relying on Internet of Things (IoT) protocols to transmit sensor readouts obtained from monitoring Physical Assets (PAs). The applications support visualization mechanisms to create Digital Twins (DTs) that mimic the behavior of the actual assets. Changes to PAs are reflected on the DTs, and vice versa. This duality enables the creation of Cognitive Digital Twins (CDTs), where the information generated by PAs is converted into knowledge at the application level. This chapter introduces a generic PA Virtualization (PAV) framework that leverages the 5-layer Internet Engineering Task Force (IETF) protocol stack by detailing an application layer smart sublayer that learns from the interaction between PAs and DTs as well as CDTs. This framework is agnostic of the CPS under consideration, regardless of whether it is Industry x.0, Connected Health, Smart City, or Building Automation. Moreover, PAV is key to supporting the training of human users interacting with CPSs through emulated PAs in the decades to come.