BTIDEC: A Novel Detection Scheme for CPU Security of Consumer Electronics
Renhai Dong, Baojiang Cui, Yi Sun, Jun Yang · IEEE Transactions on Consumer Electronics · 2023
With the rapid development of consumer electronics in Industry 5.0, the security of the CPU which is the cornerstone in the underlying hardware of the electronics has become an urgent need in sustainable and resilient technologies. Security detection of processors for multiple consumers is essential in massive consumer centric electronics in Industry 5.0. Manufacturers have proposed mitigation s at both the hardware and software levels after the CPU vulnerabilities were exposed Unfortunately many attacks against mitigation s are frequently reported. Therefore, our work is motivated to propose a detection scheme for CPU security of consumer electronics. In this paper, to address the challenges faced by consumer electronics in Industry 5.0, we put forward a novel scheme BTIDEC to i nvestigate and measure the branch target injection attacks by security analysis of BTI, speculative control flow modeling and fuzzing. The instruction sequence search space is reduced by security analysis of branch predictor and instruction sequence generation algorithm. The instruction sequence is preprocessed by speculative control flow modeling to improve the efficiency of the fuzzing module. Experiments demonstrate that BTIDEC detects the initial BTI attack on real CPUs within approximately 30 minutes, showcasing an average efficiency improvement of 57.26% over the state of the art tool. Furthermore, it achieves a high level of accuracy, identifying all known BTI sub variants on x86 microarchitectures with 98% precision, including a previously undiscovered sub variant.