Novel data encryption scheme and its application to secure multi-party communication

Abstract

Recent advances in cellular mobile telecommunication field have dramatically facilitated the multi-party collaborative work in social networks. However, the privacy issues exposed by insecure network channels and semi-trusted service providers, such as underlying data analysis and mining, have gradually aroused public concerns. In this context, a novel Multi-Party Privacy Data Encryption (MP-PDE) scheme built upon the deep learning framework is proposed. In this scheme, a four-dimensional autonomous chaotic system is initially leveraged to configurate the key-controlled cipher streams. Under the guidance of a multi-objective optimization function, the proposed encryption network manipulates the multi-party private data into a cipher image with the statistical pseudo-randomness. At the recipient side, distinct participants can decrypt the corresponding data availing their own licensing key from the identical cipher image. Furthermore, the encryption and decryption networks are equivalent except for their trainable network parameters. Finally, numerous experiments are conducted to verify the effectiveness and security of the proposed scheme.