Security enhancement of LSB-based audio steganography method

Abstract

The rapid development of digital technology has brought about the challenge of ensuring information security. Cryptography and steganography are among the various techniques available to address this challenge. These techniques come in different forms and provide reliable means of securing information and communication. Despite the existence of numerous options, new variations are still emerging. This paper concentrates on Least Significant Bit (LSB) coding, which is one of the most widely recognized and frequently utilized steganography techniques. The advantages of LSB-based coding methods are high capacity and low complexity. However, predictability is an issue. Audio files’ sizes and data redundancy make audio data perfect for steganography, as it is possible to embed large amounts of secret information and easily transmit the signal via various communication channels. However, the use of LSB coding in audio steganography has a drawback due to the extreme sensitivity of the human auditory system (HAS). This means that any noise added to the audio with data embedded in the LSBs can be detected by the HAS, which is dependent on the number of LSBs used. This work presents an improvement of the LSB-based approach, the XORing of LSBs method, with an emphasis on enhanced security. The method achieves this by randomly picking the pairs of bits to be XOR’d while preserving the characteristics of classic LSB methods. To verify the method, it has been implemented on the Cirrus Logic DSP, proving that it may be efficiently used in a real-time audio environment without interfering with HAS sensitivity levels.