Fully batch processing enabled memory integrity verification algorithm based on Merkle tree

Abstract

Memory attacks have been increasing in number recently. Adversary can manipulate memory data or break system by doing active attacks. Especially, main memory is used as a target of attack, because main memory is more vulnerable than other components, such as CPU. To prevent adversary’s active attack, memory integrity verification algorithm has been proposed. Protection of computer’s memory integrity is important in situations where attacks on the computer systems are a threat. As technology has advanced, computer systems migrate from wire-based to wireless system. A lot of memory integrity verification algorithms are already developed, but these algorithms do not consider new wireless platform. Wireless platform is constrained by a lack of storage and power supply in comparison with wire-based system, therefore computational overhead and storage overhead must be considered when applying to algorithm, which is used in wireless system. In this study, integrity verification performance can be improved by doing batch-processing. Previous verification algorithms based on Merkle tree do not support fully batch processing verification. We propose fully batch processing enabled memory integrity verification algorithm based on Merkle tree. This algorithms can verify memory integrity in completely batches. For implement our algorithm, we use Incremental multiset hash function, and as a result, consume only 480-bit on-chip storage. Reducing consumption of on-chip storage leads to improving on the performance of computation. We implement our algorithm and previous memory integrity verification algorithms based on standard Merkle tree and lazy-processing Merkle tree in simulator to compare their performance. Our algorithm offers better system performance overall, especially when attack rarely occur.