Predicting IoT attacks

La Trobe researchers have developed a new security framework designed to predict and detect cybersecurity attacks on Internet of Things (IoT) devices.

La Trobe researchers have developed a new security framework designed to predict and detect cybersecurity attacks on Internet of Things (IoT) devices.

“IoT refers to the large network of devices that connect and share data with each other. The number of IoT devices is growing rapidly, connecting everything from homes to industries. However, with this growth comes significant cybersecurity challenges," explains lead researcher, Dr Rudri Kalaria.

"Many IoT devices are vulnerable to attacks like unauthorised access and data breaches due to insufficient security measures. As IoT becomes increasingly integral to our lives, prioritising robust security practices is critical to safeguarding both personal and organisational data.”

To address this, the research team – ASM Kayes, Wenny Rahayu, Eric Pardede, and Ahmad Salehi Shahraki – has developed a new security framework called the IoTPredictor.

“The IoTPredictor not only identifies potential threats but also forecasts device behaviours, allowing for proactive security measures,” Dr Kalaria says.

“It shifts the focus from traditional, reactive security measures to a proactive model that can predict and mitigate threats before they escalate. This means it is more robust against ever-evolving cyber threats.”

“We hope this framework will enhance cybersecurity in critical areas such as smart cities, healthcare and industrial systems. This means that IoTPredictor has the potential to provide long-term benefits for both industry and public safety.”

The research team have published a paper evaluating the effectiveness of the IoTPredictor and will soon begin extensive field tests.

“Our initial experiments show the IoTPredictor is effective in predicting and preventing malicious activities. We will now conduct field tests to confirm this under different conditions.”

“This will provide valuable insights into how the model performs in real-world scenarios, helping to refine its predictive accuracy and adaptability to new types of attacks.”