Broad adoption of AI is critical for Canada to build a smart manufacturing ecosystem, but few Canadian industrial companies have actually deployed at least one AI project in their operations. At the same time, cybersecurity has become a pressing issue for industrials. Recent events have shown the detrimental effects of cyberattacks on a company’s operations and their potential consequences on the overall economy that depends on it. Considering that security breaches are most common at the edges of the network, the challenge is even more complex for manufacturers with assets dispatched all over the world. How can a company protect its assets and ensure the good functioning of their industrial systems and machines from a central monitoring centre in such a context? Can artificial intelligence and cloud technologies provide solutions to effectively manage cybersecurity challenges in an industrial environment?
BI Expertise and its BX Lab are specialized in collaborative research and innovation, and develop specific hardware and software solutions leveraging machine learning and cloud technologies. One of their manufacturing clients builds recycling machines that they ship to recycling facilities worldwide. Remotely monitoring their machines for maintenance purposes and updating their systems represents a major challenge: How can the machine securely connect with the central system when necessary? How can we make sure the parties connecting to the system for maintenance operations are authorized and do so securely? etc. In 2020, BI Expertise partnered with their manufacturing client to test a pilot project and tackle the issue of remotely securing their assets (using cryptography), maintaining and updating the systems of distributed assets from their headquarters, without having a technician on premise near the remote machine. To overcome the major obstacles of realizing such a task, the project made use of functionalities that the cloud allows to deploy and of machine learning techniques to monitor the security of the system.
In this case study on cybersecurity in an Industry 4.0 context, we will guide cybersecurity and maintenance managers in industrial companies with distributed assets on how 1) we leveraged the cloud for the dynamic encryption of the machine’s hard drive to avoid storing encryption keys on unsecured locations; 2) we integrated multiple cloud instances to deploy a main control centre and a dedicated machine learning environment; and 3) automated the training and validation of machine learning algorithms to 4) allow for the automation of remote machines’ system update. This presentation will then discuss how such a project opens the way to the broader adoption of machine learning techniques, especially unsupervised methods, in the field of cybersecurity and the Industrial Internet of Things (IIoT).