Crypto-Deep Reinforcement Learning Based Cloud Security For Trusted Communication

One of the most essential business models in modern information technology is cloud computing. It offers a variety of services for user interaction as well as low-cost hardware and software. Cloud services are built on newline virtualization designs that use multi-tenancy for improved resource management and newline strong isolation across several Virtual Machines (VMs). Despite advancements in virtualization, data security and isolation assurances continue to be the major difficulties faced by cloud providers using existing approaches. To overcome this problem, Deep Reinforcement Learning is applied to offload the task and also to detect the generalized attackers in the cloud network. This proposed solution enables remote data monitoring approaches such as identity-based linear classification algorithms for VM attack classification channels. It can minimise data secrecy and increase communication by using a reinforcement learning technique. The attacker channel is identified using identity-based linear classification when data is transferred/retrieved from the VM cloud server. When the classifier finds channel misbehaviour, the port or channel may be blocked, and the communication of other accessible ports may be modified maintaining the end to end communication secrecy using the improved Multi Agent Deep Reinforcement Learning (MADRL). The service verification is done to ensure that users have secure access to the cloud server. When an unknown request to the cloud server runs the key authentication to check the user authorization, this linear classification trains the existingside-channel attack datasets to the classifier and detects the VM cloud's attack channel. In terms of overall performance, the proposed methodology is investigated and compared to the existing approaches.