Abstract

In Neural Cryptography, Artificial Neural Networks are used for the process of key generation and encryption. Tree Parity Machine (TPM) is a single layer neural network that approaches symmetric key exchange using the process of mutual learning. This method is exploited to design a secure key exchange protocol, where the sender and the receiver TPMs are synchronized to obtain an identically tuned weight vectors in both the networks. The synchronized TPMs are then capable of generating a key stream. The time required for synchronization depends on the initial weight vectors which are randomly initialized. In the proposed method, the process of synchronization is expedited using Cuckoo Search (CS) Algorithm used for the generation of optimal weights.