Scholarlink

<p>The given paper introduces a cloud security model that utilizes the virtual-Berkeley open infrastructure for network computing (V-BOINC) system to address the privacy issues in cloud computing by integrating steganography and deep learning so that data can be made secured in ad-hoc cloud. The major contributions are the design of a two-phase system in which the first phase is to define flexible ad-hoc cloud architecture and the second phase is to design a secure steganography scheme, assisted with deep learning algorithms. The proposed framework provides enhanced data security against a variety of threat vectors compared to conventional systems with low computational cost. The model is empirically verified based on efficiency, reliability, and data integrity. It should be noted that this approach achieves extremely high performance in hiding and transferring encrypted information and images, with good improvements in attack resistance and operation stability. The results show that the proposed system has the highest success rate (99.8%), highest accuracy (99.9%), highest encryption efficiency (99.9%), highest transmission reliability (98%) and highest attack resilience (96%).</p>

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

Deep learning-based steganography framework to enhance ad-hoc cloud security

  • 期刊论文

  • 发表日期：

    2025年7月26日

查看全文
  • Vikas Lamba,
  • Jyoti Goyal,
  • Ramesh N. Koppar,
  • Kumar Puttaswamy Gowda,
  • Mallareddy Adudhodla,
  • Brajesh Kumar Singh,
  • Gurwinder Singh

摘要

The given paper introduces a cloud security model that utilizes the virtual-Berkeley open infrastructure for network computing (V-BOINC) system to address the privacy issues in cloud computing by integrating steganography and deep learning so that data can be made secured in ad-hoc cloud. The major contributions are the design of a two-phase system in which the first phase is to define flexible ad-hoc cloud architecture and the second phase is to design a secure steganography scheme, assisted with deep learning algorithms. The proposed framework provides enhanced data security against a variety of threat vectors compared to conventional systems with low computational cost. The model is empirically verified based on efficiency, reliability, and data integrity. It should be noted that this approach achieves extremely high performance in hiding and transferring encrypted information and images, with good improvements in attack resistance and operation stability. The results show that the proposed system has the highest success rate (99.8%), highest accuracy (99.9%), highest encryption efficiency (99.9%), highest transmission reliability (98%) and highest attack resilience (96%).