IIC Journal of Innovation 6th Edition | Page 24

Industrial Intelligence: AI’s Implications on Security, Seamlessness and Services for the IIoT welcomed into everyday life. The result is a virtuous cycle where AI has gained consumer trust and business interest that has led to a snowballing growth. are raised. AI may be used to develop context-aware security systems and alarms for IoT devices. This AI leverages the fact that most IoT devices are mirrored in the Cloud or another central repository with scalable computing. Physical system behavior is observed, with sparse data being sent to a digital duplicate using a model to “interpolate” these data into a rich representation. This model starts out as generalized by object type, but over time adapts to the particular nuances of the mirrored object’s own sensors, environment, and use cases. These models interact with one another in the Cloud so that they learn their place in a larger system. T HREE C ASES FOR A RTIFICIAL I NTELLIGENCE Just as in the consumer space, industry has seen the benefit of AI as an effective force (revenue) multiplier. Corporations have visions of synthesizing AI and IoT to improve operational efficiency while reducing cost and to provide process analytics, though the biggest opportunities have yet to be explored. We present a non-exhaustive list of three areas where AI can transform the IoT, and in turn industry: “Cognitive Firewalls” and “Cognitive Supervisors” use each model to evaluate the impact of commands to ensure they are benign prior to execution, or to identify when a process behaves anomalously. 2 The Firewall uses the adaptively-learned model to “test” a command digitally to ensure it does not violate any known or learned limits prior to forwarding it to the related physical device. For example, the Cognitive Firewall could be used to protect a robot arm from malicious commands. When the arm is sent commands that cause the arm’s mirror to intersect with a second robotic arm in the same Cloud Factory, the command is rejected. Similarly, an oven may learn that it is located within a food processing facility and that its purpose is to bake cookies, so a safe maximum temperature limit is 500F. 1. Context-aware security 2. Building seamless “human” IoT interactions 3. Intelligent services Security IoT relies on the ability to collect data and control systems at scale. Collection and control demand a high level of practitioner and end-user trust. Trusting users are more likely to share valuable data, while trusting companies are more likely to use IoT to control core operations where analytics may unlock the biggest efficiency improvements. Creating systems that can adequately secure equipment from intentional and accidental malicious use is challenging. When sensitive actuators such as power infrastructure or connected vehicles join the mix, the stakes 2 The Cognitive Supervisor may be used to identify process problems. For example, the Supervisor may learn that a particular smelting furnace at 80% duty cycle reaches a The Future Internet of Things: Secure, Efficient, and Model-Based http://ieeexplore.ieee.org/document/8055694/ IIC Journal of Innovation - 23 -