Research And Application Of Privacy Protection For Distributed Compressive Video Sensing

The booming development of Internet of Things has led to the emergence of novel application fields such as wireless multimedia sensor networks,cloud storage and edge computing.The data from hundreds of millions of individuals across the world could be no longer limited to the scalar information,but more to the multimedia content such as images and videos.As a novel video coding technology,distributed compressive video sensing(DCVS)breaks through the constraints of traditional video coding by means of independent encoding and joint decoding.Besides,the privacy of increasing video data in encoding and transmission also needs to be strongly protected.Thus,the research and application of privacy protection for DCVS in this thesis mainly includes:(1)The scheme applied in the wireless multimedia sensor networks environment is designed,which not only can provide low-cost and high-efficiency of privacy protection for the video content,but also improves the encoding efficiency of DCVS.At First,the two-pattern adaptive group of pictures selection is adopted to mark whether each frame of the input video is the key frame or the non-key frame.Next,the deterministic binary block diagonal measurement matrix is optimized for sampling every frame to reduce the encoding complexity.As for the privacy and security issue of the video information,the pseudo-random number generator is adopted to generate keys and then the scramblingsubstitution-diffusion encryption algorithm is designed to resist four typical cryptgraphy attacks.The experiments demonstrate that this scheme not only can guarantee efficiency and security of video coding and transmission,but also meets the demands of real-time transmission for videos in wireless multimedia sensor networks.(2)The novel hierarchical edge computing architecture is designed by making full use of three modes(multi-access edge computing,cloudlets and fog computing)of edge computing.And the scheme of privacy protection for big video data storage based on this architecture with low complexity and high security is proposed.Divide a video into three parts and then store them into completely different facilities.Specifically,the most significant bits of a key frame are stored directly in local sensor devices while the least significant bits of that are encrypted and then sent to the semi-trusted cloudlets.Besides,the difference between a non-key frame and a key frame of the same group is sampled by the two-layer parallel compressed sensing firstly,encrypted by the 2D Logistic-Skew?Tent map secondly,transmitted to the cloud server at last.Simulation experiments and theoretical analyses show that this scheme not only provides smart privacy protection effectively for big video data storage based on hierarchical edge computing,but also avoids additional computation burden and storage pressure as well.