Unique and Random Key Generation Using Deep Convolutional Neural Network and Genetic Algorithm for Secure Data Communication Over Wireless Network
S. Venkatesan, M. Ramakrishnan, Manne Archana · 2023
Wireless network plays a dynamic role in the high-speed communication of data across the globe. Achieving highly secure data communication and protecting the confidential data from illegal access are the key concerns in the wireless network. Encryption and decryption ensure better data security by using public key cryptography. However, the public key could be easily hacked by unauthorized user. This makes the network vulnerable to the attacks. To ensure improved data security, there is a need to create encryption key and decryption key that are unique in nature, diverse and random. In the proposed work, the cryptographic keys are generated by using Elliptic Curve Diffie-Hellman (ECDH) with Deep Convolutional Neural Network (DCNN) and Genetic Algorithm (GA). High data confidentiality and data integrity are achieved by preventing unauthorized manipulations and denial of the message. The GA generates a population having maximum fitness value, which acts as the transitional cipher text applied as input to the DCNN to encrypt the original message. The DCNN uses its security key in the form of weights, by using the backpropagation algorithm. The ECDH algorithm generates unique, encrypted messages by using the DCNN keys that cannot be accessed deprived of authorization. The keys possess the randomness essential for both security and key strength. Hence, the proposed work improves data security by evaluating the performance of DCNNs and GAs on the user data over the elliptic curves with high accuracy.