A hybrid AES encryption for IOT using Adversarial Network

AES is one of the most commonly used methods to encrypt data for Internet of things(IOT). However, the encryption process usually complies with determined rules and fixed keys, once the keys are exposed to the attackers, they can effectively carry out correlation analysis attacks by reverse decryption. In this work, we proposed to integrate Generative adversarial Network (GAN) into inner layer of AES for better encryption. As the GAN generated encryption is not based on determined rules and it needs learning process to encoding and decoding. Therefore, it provides more security facing reverse cracking. Experiments prove the effectiveness of our proposal.