A group communication needs high security in cloud due to the security issues like group forward and backward secrecy. Many works have addressed those problems, but still they have some lack in their solutions. To address this issue, we propose a new deep learning-based key transmission (DLKT) protocol. This protocol gives the secure group communication in cloud by using autoencoder-based generative adversarial network framework which allows users to transmit their key or message in a secure manner. Both hash and reverse hash techniques are used in this protocol to perform secure session establishment process. A novelty is making by applying deep learning technique into the key transmission protocol for improving the efficiency and security level of the proposed method. The outcome of the work proves that our DLKT protocol reduces the operation time, increases the attack detection rate and accuracy of attack detection compared with existing works and provides very high security.

由于群前向和后向保密等安全问题，群通信在云中需要高安全性。许多工作已经解决了这些问题，但仍然缺乏解决方案。为了解决这个问题，我们提出了一种新的基于深度学习的密钥传输（DLKT）协议。该协议通过使用基于自动编码器的生成对抗网络框架在云中提供安全的群组通信，该框架允许用户以安全的方式传输他们的密钥或消息。该协议使用散列和反向散列技术来执行安全会话建立过程。通过将深度学习技术应用于密钥传输协议，以提高所提出方法的效率和安全级别，这是一种新颖性。