New AI-Powered Chip Security System Detects Threats in Real Time
A new security breakthrough could change how modern Systems-on-Chip (SoC) are protected. Researchers have developed an Intelligent Hardware Monitoring System (IMS) that detects AXI protocol violations in real time. This innovation moves beyond traditional access controls by analysing protocol behaviour directly.
The IMS was designed to tackle a rising threat: vulnerabilities in the Advanced eXtensible Interface (AXI) protocol. These flaws can lead to denial-of-service attacks, disrupting critical systems. To address this, the team built a neural network-based system capable of real-time semantic analysis.
Testing took place on a Zynq UltraScale+ MPSoC ZCU104 board. The system showed a tiny hardware footprint and had almost no effect on the design’s operating frequency. It was also integrated into a RISC-V SoC as a memory-mapped IP core, proving its practical use in standard architectures. Performance results were impressive, with 98.7% detection accuracy and minimal latency overhead. The researchers further supported transparency by releasing their AXI bus header attack dataset. This allows other experts to reproduce the work and explore new security solutions.
The IMS introduces a shift in SoC security, focusing on real-time protocol monitoring instead of just access control. Its high accuracy and low resource demands make it suitable for edge environments. The public dataset also enables further research into stronger defences against evolving threats.
