Scholarlink

This study investigates the performance of GPGPU-accelerated file carving tools, with a focus on the development and evaluation of a new tool, Aletheia. The research specifically examines the use of the PFAC! (PFAC!) string search algorithm in both CPU and GPU environments, comparing it to existing tools like OpenForensics, Foremost, and Scalpel. The experimental setup utilized two distinct environments: a personal desktop with an i5 3570k CPU and GTX 980 Ti GPU, and an AWS g3s.xlarge instance with a Xeon E5-2686 v4 CPU and TESLA M60 GPU. Empirical results show that Aletheia achieves competitive performance in file carving, demonstrating efficient pattern searching across multiple file types, including PNG, JPG, GIF, PDF, DOC, and HTML. For a 400 MB! (MB!) file system, Aletheia’s performance was comparable to OpenForensics, with search times of approximately 24.512 s for a 2 2GB! (GB!) file system in the AWS environment. The tool’s effectiveness was further validated by addressing issues such as incorrect memory reads in HTML file types and optimizing performance through RAM disks and different file chunk sizes. In conclusion, the development of Aletheia confirms the viability of GPGPU methods for accelerating file carving tasks. The tool’s design and implementation, leveraging PFAC! and CUDA, show significant potential for improving the speed and accuracy of digital forensic searches. The findings suggest that while GPGPU-accelerated tools offer substantial performance benefits, practical considerations such as hardware limitations and compatibility issues must be addressed to optimize their deployment in real-world scenarios.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Accelerating Digital Forensics with GPGPU Techniques: A Performance Evaluation

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ritesh Mukherjee

摘要

This study investigates the performance of GPGPU-accelerated file carving tools, with a focus on the development and evaluation of a new tool, Aletheia. The research specifically examines the use of the PFAC! (PFAC!) string search algorithm in both CPU and GPU environments, comparing it to existing tools like OpenForensics, Foremost, and Scalpel. The experimental setup utilized two distinct environments: a personal desktop with an i5 3570k CPU and GTX 980 Ti GPU, and an AWS g3s.xlarge instance with a Xeon E5-2686 v4 CPU and TESLA M60 GPU. Empirical results show that Aletheia achieves competitive performance in file carving, demonstrating efficient pattern searching across multiple file types, including PNG, JPG, GIF, PDF, DOC, and HTML. For a 400 MB! (MB!) file system, Aletheia’s performance was comparable to OpenForensics, with search times of approximately 24.512 s for a 2 2GB! (GB!) file system in the AWS environment. The tool’s effectiveness was further validated by addressing issues such as incorrect memory reads in HTML file types and optimizing performance through RAM disks and different file chunk sizes. In conclusion, the development of Aletheia confirms the viability of GPGPU methods for accelerating file carving tasks. The tool’s design and implementation, leveraging PFAC! and CUDA, show significant potential for improving the speed and accuracy of digital forensic searches. The findings suggest that while GPGPU-accelerated tools offer substantial performance benefits, practical considerations such as hardware limitations and compatibility issues must be addressed to optimize their deployment in real-world scenarios.