An Integrity Protection Scheme Based on Binary Multi-granularity Incremental Tree for Embedded Computing Platform

Data integrity is an effective approach to protect the off-chip memory from active attacks such as data modification, injection and malicious code execution. However, in the existing integrity checking schemes, constructing the checking tree and operating massive computation occupied excessive memory space during checking the integrity of the memory data, and those schemes cannot be directly applied to the embedded system due to the limited system resources. A memory integrity protection scheme based on a binary multi-granularity incremental hash tree (MIT) was proposed. By applying this scheme, there are three improvements: reducing the scale of the node of the checking tree, optimizing the structure of the checking tree and accelerating the verification rate. The proposed memory integrity protection scheme was quantitatively evaluated by the SimpleScalar architecture simulation tool. The experimental results show that, compared with the traditional hash method, the rate of the overhead memory occupation and the average system performance are decreased by 48.72% and 32.62% respectively, which have better performance than the existing methods.