Hybrid Cryptography and Steganography-Based Security System for IoT Networks

Abstract

Despite the fact that many IoT devices are publicly accessible to everyone on the network, understanding the security risks and threats posed by cyber attacks is critical; as a result, it should be safeguarded. Plain text is constructed into encrypted text, before being delivered by using cryptography, and is then reconstructed back to plain text after receiving a response from the recipient. The steganography technique can be used to hide sensitive information incorporated in a text, audio, or video file. One approach is to hide data in bits that correspond to successive rows of pixels with the same color in an image file.  As a consequence, the image file retains the original's appearance while also containing "noise" patterns made out of common, unencrypted data. To do this, the encrypted data is subtly applied to the redundant data. In this work, it is suggested that IoT network data be encrypted using cryptography, and that an encrypted message be concealed inside an image file using steganography. Additionally, it is suggested to enhance the number of bits that may be stored within a single picture pixel.  The payload that may be sent through an image is significantly increased by incorporating Convolutional Neural Networks into the classic steganography technique. In this work, we propose, design, and train Convolutional Neural Networks (CNN) to enhance the amount of data that can be securely encrypted and decrypted to show the original message.