Logo: to the web site of Uppsala University

While the Internet of Things (IoT), Wireless Sensor Networks (WSNs), and the techniques for extracting digital data have seen continuous advancements, so does the cyber-threat landscape. Virtual sensors which normally use abstraction layers that operate over a physical infrastructure to achieve their objectives, have seen rapid adoption, for example, it has aided in achieving manufacturing 4.0. This abstract layer has in the recent past seen tremendous proliferation within the sensor-based platform. Coupled with data pre-processing and key compliance with the guidelines for information security, incident investigation principles, and processes. This paper discusses a step towards a Learning-enabled (LE) Virtual Sensor Forensic (VSF) architecture that is compliant with edge intelligence technology, which is based on an initially proposed generic VSF architecture. Furthermore, apart from the learning capabilities, the LE-VSF architecture considers proactive and reactive investigation techniques by assuming an Internet of Vehicle (IoV) attack scenario, enhancing the reliability of a forensically sound data source. This proposition is essential in any sensor-based abstraction where the forensic analysis would otherwise be cumbersome and susceptible to noise.