Improvement of the Efficiency of Neural Cryptography-Based Secret Key Exchange Algorithm

Owing to new security threats and relevant changes in network computing environments, key exchange methods that replace conventional public key exchange algorithms are being researched. The neural cryptography-based key exchange algorithm proposed recently uses neural synchronization as an alternative to public-key methods. However, the learning-based synchronization method uses considerable communication resources to generate output values and share the results. The efficiency of this method depends on the type of algorithm and is affected by both the communication rounds for exchanging output values and the number of weight learning rounds. To improve its efficiency, this paper proposes 1-h random walk and batch scheme methods and verifies their efficiency and security.