Password recovery for ZIP files based on ARM-FPGA cluster

Abstract

Password recovery of ZIP encrypted files is an important problem in computer forensics. The encryption is based on standard cryptographic algorithms as SHA1, HMAC and AES. The traditional methods such as dictionary and brute-force require very large computing power and techniques of reducing the password space. In this paper, we have developed a distributed password recovery system based on Zynq (a heterogeneous chip combining ARM CPU and FPGA fabric) cluster. The FPGA provides hardware acceleration for cryptographic algorithms. And the ARM completes the decompression after decryption to check candidate passwords. To reduce the computation of unzip, we only decompress the header of the file compressed according to different headers of common document formats. We adopt a cluster-building methodology to improve parallelism and calculation power. Finally, the experimental results show that single node is as fast as a core of i7-3770 CPU and the 48-node cluster can check 50,000 passwords per second. It also achieves about 2 × energy efficiency.