Research On Data Privacy Preservation Based On Compressive Sensing

In the era of interconnection of all things,thousands of sensors,servers and intelligent terminals constitute a more extensive Internet of things than the traditional Internet.Information interaction is no longer limited to people,Human beings can perceive information from the nature.With the development of the Internet of things and with the promotion of the commercial use of 5G,the number of Internet terminals will further increase,which serve us in many areas,such as health monitoring,intelligent traffic control,crime prevention,intelligent electricity meter measurement,etc.Then,because of the particularity of the hardware environment and the limitation of objective conditions,the sensing devices often face the resource-limited problems,including the limited storage space,the limited computing capacity,the limited power,the limited size and so on.For instance,in the health monitoring network,the sensing devices are beginning to lightweight,even Implantable for the sake of pursuing the ultimate user experience.At this time,the traditional cryptographic techniques with high complexity are no longer applicable such that the security of the sampling data can not be guaranteed.The cryptographic technology based on compressive sensing,called as compressive sensing cipher,can be used as a lightweight cryptosystem built in the sensing layer,providing the first security layer for data at nearly zero cost,so it is often advocated to be used in the resource-limitted scenarios.However,the linearity of compressive sensing determines the security vulnerability of compressive sensing cipher.In this paper,the privacy-preserving compressive sensing is studied as follows:(1)An improved version of the compressive sensing encryption algorithm is proposed.Firstly,the sampling data are divided into sensitive data and nonsensitive data and then transmitted respectively.In addition,a cloud-driven secure processing scheme for compressive sensing data is proposed,in which the sensitive data are small but vital and its security is guaranteed by the proposed chaotic encryption algorithm and the hash algorithm which can fix the position of the tampered data;the nonsensitive data are of equal magnitude with the sampling data and can not leak any valuable information,so them are directly transmitted over the public channel;high-complexity signal reconstruction algorithm is implemented in the cloud servers.(2)An image sharing scheme based on compressed sensing is proposed,which combines the chaos-based compressive sensing encryprion technique and the polynomialbased secret sharing scheme.The sampling data are divided into several shadows and then shared between a group of participants to ensure the high security in a low-complexity way.This “1 + 1 > 2” combination scheme not only improves the confidentiality of data,but also get many new functions which do not exist in the existing image sharing schemes,including pixel compression,progressive preview and double robustness.Most importantly,in the face of the identity attack to the shared group,the proposed scheme innovatively establishes a trust mechanism,which can not only detect the cheating behavior but also identify the cheater.