面向网络流媒体隐蔽通信的非均衡隐写方法研究

Multidimensional fusion steganography is a novel idea recently proposed for covert communications over network streaming media, which aims to achieve information hiding by comprehensively utilizing steganographic characteristics on all dimensions (including streaming media signals and network protocols). However, its performance is greatly challenged by the imbalance in steganographic transparencies of different dimensions. Therefore, in this project, we are motivated to research unequal multidimensional steganography from the view of unbalanced protection to achieve high-performance covert communications. The main research contents include: First, the steganographic transparency of each hiding position on all dimensions will be exactly evaluated and quantized, and a classified cover characteristic database for each dimension will be constructed. Second, classified steganographic methods will be researched respectively from the views of humidity degree model and graded dynamic encoding. Third, a model for unequal multidimensional fusion covert communications (UMFCC) will be constructed, and related fusion strategies will be designed. Fourth, the synchronization and reliable transmission mechanisms for UMFCC will be accordingly established. Lastly, using the above- mentioned techniques, a prototype UMFCC system based on Voice-over-IP will be designed and implemented. The research findings of this project will further enrich the theory of information hiding and covert communication, provide a fundamental basis and a practical guide for secure communications of secret information over public networks, and thereby have significant application value.

综合利用多个维度上（即流媒体信号和各层协议维度）的隐藏特性以实现信息隐藏是近年来网络流媒体隐蔽通信研究中提出的一种新思路，称为多维融合隐写。然而，各维度载体隐藏透明性的不均衡给多维隐藏的性能构成了极大的挑战。针对这一问题，本项目拟从"不等保护"的角度研究非均衡多维隐写技术以实现高性能隐蔽通信，主要内容包括：①准确评价并量化各维度上所有可隐藏位置的隐藏透明性，建立各维度上的分级载体特征库；②从湿度级模型和分级动态编码两个角度研究各维度上的分级隐写方法；③建立基于网络流媒体的非均衡多维隐蔽通信模型，并设计实现相应的融合策略；④建立非均衡多维融合隐蔽通信下的同步及可靠传输机制；⑤综合应用上述成果，设计并实现基于VoIP的非均衡多维融合隐蔽通信原型系统。本项目的研究成果将进一步丰富信息隐藏和隐蔽通信理论体系，为实现隐秘信息在公共网络中的安全通信提供理论依据和实践指导，具有重要的应用价值。

本项目从分析网络流媒体隐写不均衡性着手，借鉴信息编码理论，深入研究了面向以IP语音为代表网络流媒体的分级隐写和多维隐藏模式下的非均衡隐写理论和方法，取得了系列研究成果。概括而言，已完成的主要工作包括：（1）从客观的语音质量感知、信噪比、相似度和抗隐写分析等方面对载体隐写特性进行了评价和量化；（2）从湿度级模型和分级动态编码两个角度提出了多种分级隐写方法，如基于可量化性能分级的自适应隐写方法、基于矩阵随机扩展的改进湿纸码模型、基于随机二进制矩阵的高效隐写方法、基于（2n-1, 2n）编码的隐写方法等；（3）提出了多种非均衡多维隐蔽通信策略, 如基于帧比特率变换的隐写、基于自适应速率调制的隐写、可调节隐写编码方法以及基于组合优化的多编码分组最优化融合策略、非均衡隐写策略等;（4）建立了与非均衡多维融合隐蔽通信相适应的同步及可靠传输机制；（5）以IP语音为代表载体，实现了上述隐写方法，并对其进行了测试和评估。上述研究成果共发表学术论文12篇，其中SCI检索7篇，EI检索4篇；申请并获得授权发明专利3项、软件著作权3项。本项目的研究成果进一步丰富了信息隐藏和隐蔽通信理论体系，并可为实现隐秘信息在公共网络中的安全通信提供理论依据和实践指导，具有重要的应用价值。