Working to Detect Attacks in Real Time
Wednesday, September 7, 2016 @ 04:09 PM gHale
In a move to boost productivity by spotting a glitch on the factory floor in real time and reconfiguring around it, “a revolutionary methodology for controlling manufacturing systems” is now in development.
The venture, a $4 million project led by University of Michigan engineering researchers and others from the University of Illinois and Cornell University, aims to increase factory productivity and American competitiveness.
Modern advanced manufacturing plants hold hundreds of software and hardware components. Robots, conveyer belts, sensors, control systems and communication networks have intricately choreographed roles in a sector that yields 12 percent of the nation’s GDP.
Machine failures, operators’ mistakes and, cyberattacks can halt production, leading to expensive unscheduled downtime or potentially dangerous situations.
One case in point occurred in 2015, when hackers breached the control system in a steel mill in Germany and made it impossible for operators to properly shut down a blast furnace. The damage was massive.
The new project — from a grant from the National Science Foundation — will yield what the researchers said is “a revolutionary methodology for controlling manufacturing systems.” They call the methodology “software-defined control.”
Central to the new approach is a continuous, full simulation of the manufacturing plant. The team will produce this simulation and develop software that compares a plant’s actual operation to what they’d expect based on the simulation.
“The idea is you have the physical manufacturing plant and the simulated model of the plant so if there’s a difference between the two, you can detect a fault or a cyber-intrusion,” said project principal investigator Dawn Tilbury, associate dean for research and professor of mechanical engineering at the U-M College of Engineering. “The goal is to develop control systems for manufacturing systems that are secure and reconfigurable automatically.”
Such systems could reprogram how parts flow through the plant to avoid a faulty piece of equipment.
“Our work aims to develop the science and enabling technologies to transform manufacturing systems from the current paradigm of low efficiency and high susceptibility to system disruptions to a new era of system-level anomaly detection, classification and action,” said Kira Barton, U-M assistant professor of mechanical engineering. “This will lead to less downtime, faster responses to disruptions and a more efficient manufacturing system.”
The rise of automation calls for a better way to keep tabs on plant operation, Tilbury said.
“Automation may increase efficiency and raise quality, but it brings with it vulnerabilities,” she said.
Robots are networked, and the companies that produced them can often log in remotely to make repairs. These are legitimate endeavors, but they can also be weak links in the cybersecurity of the system.
Susceptible to Disruptions
As manufacturing systems become more complex and digitally connected, they become increasingly susceptible to disruptions that can cause significant financial losses.
The U.S. Department of Homeland Security investigated 97 cyberattacks at critical manufacturers during the fiscal year ending in June 2015, according to a report by its Industrial Control Systems Cybersecurity Emergency Response Team.
Cyberattacks to manufacturing plants are a relatively new phenomenon. The FY 2015 number federal officials examined was double the previous year. A clear and present risk — and this one to the nation’s economy — is of unscheduled factory downtime. It’s one of the most prevalent causes of inefficiency in manufacturing, the researchers said. When a cyberattack or a broken machine stops a factory in its tracks, the cost can run tens of thousands of dollars a minute.
“Manufacturing systems produce the vast majority of products in the modern world — automobiles, computers, textiles and toys, to name just a few,” said Sibin Mohan, U of I research assistant professor of computer science. “Improvements in the design and operation of manufacturing systems will have a huge financial and social impact on companies and consumers, and that’s crucial to economic competitiveness.”
While the project focuses on discrete part manufacturing, the researchers said it translates well to semiconductor manufacturing and batch processes.
“The work will combine techniques from multiple disciplines, spanning control theory, modeling of physical properties, machine learning, and cybersecurity, to detect and deter such attacks. The collaboration will be exciting as well as challenging,” said Elaine Shi, associate professor of computer science at Cornell.