The Industrial Internet of Things, or IIoT, has recently gained popularity. Powered by wireless 5G connectivity and artificial intelligence (AI), IIoT holds the ability to analyze critical problems and provide solutions that can improve the operational performance of industries ranging from manufacturing to healthcare.
IoT is highly user-centric — it connects TVs, voice assistants, refrigerators, etc. — whereas IIoT deals with enhancing the health, safety, or efficiency of larger systems, bridging hardware with software, and carrying out data analysis to provide real-time insights.
However, while IIoT does have many advantages, it also comes with its share of vulnerabilities such as security threats in the form of attacks trying to disturb the network or siphoning resources. As IIoT is getting more popular in industries, it is becoming crucial to develop an efficient system to handle such security concerns. So, a team of multinational researchers led by Prof. Gwanggil Jeon from Incheon National University stepped up to this challenge.
They took a deep dive into the world of 5G-enabled IIoT to explore its threats and come up with a novel solution to the problem. In a recent review published online in IEEE Transactions on Industrial Informatics, the team presented an AI- and deep learning-based malware detection system for 5G-assisted IIoT systems.
Prof. Jeon explains the rationale behind the study: “Security threats can often lead to operation or deployment failure in IIoT systems, which can create high-risk situations. So, we decided to investigate and compare available research, find out the gaps, and propose a new design for a security system that can not only detect malware attacks in IIoT systems, but also classify them.”
The system developed by the team uses a method called grayscale image visualization with a deep learning network for analyzing the malware, and further applies a multi-level convolutional neural network (CNN) architecture to categorize the malware attack into different types. The team also integrated this security system with 5G, which allows for low latency and high throughput sharing of real-time data and diagnostics.
Compared to conventional system architectures, the new design showed an improved accuracy that reached 97 percent on the benchmark dataset. They also discovered that the reason behind such high accuracy is the system’s ability to extract complementary discriminative features by combining multiple layers of information.
This new malware classification system can be used to secure real-time connectivity applications such as smart cities and autonomous vehicles. It also provides solid groundwork for the development of advanced security systems that can curb a wide range of cybercrime activities.
