Multi-Layered Security Framework Combining Steganography and DNA Coding

Abstract

With the rapid expansion of digital communication and data sharing, ensuring robust security for sensitive information has become increasingly critical, particularly when data are transmitted over public networks. Traditional encryption techniques are increasingly vulnerable to evolving cyber threats, making single-layer security mechanisms less effective. This study proposes a multi-layered security approach that integrates cryptographic and steganographic techniques to enhance data protection. The framework leverages advanced methods such as encrypted data embedding in images, DNA sequence coding, QR codes, and least significant bit (LSB) steganography. To evaluate its effectiveness, experiments were conducted using text messages, text files, and images, with security assessments based on PSNR, MSE, SNR, and encryption–decryption times for text data. Image security was analyzed through visual inspection, correlation, entropy, standard deviation, key space analysis, randomness, and differential analysis. The proposed method demonstrated strong resilience against differential cryptanalysis, achieving high NPCR values (99.5784%, 99.4292%, and 99.5784%) and UACI values (33.5873%, 33.5149%, and 33.3745%), indicating robust diffusion and confusion properties. These results highlight the reliability and effectiveness of the proposed framework in safeguarding data integrity and confidentiality, providing a promising direction for future cryptographic research.

1. Introduction

The internet and communication applications have brought various advantages over the past two decades, enabling virtual conferences, real-time data exchange, and seamless connectivity across industries such as social media, finance, and government services [1,2]. With the increasing reliance on digital platforms, securing data transmissions has become a critical challenge. Sensitive information, including financial records, personal communications, and confidential corporate data, is frequently targeted by cybercriminals employing sophisticated attack techniques [3,4]. Reports indicate a significant rise in cyber threats such as data breaches, phishing, and ransomware attacks, highlighting the urgent need for robust security mechanisms [5,6,7].

By employing encryption techniques to ensure secrecy, integrity, and authenticity, cryptography plays a crucial part in protecting online communications [8,9]. It is frequently utilized in many different applications, such as secure chat systems, mobile communications, banking transactions, and e-commerce. However, its efficacy, encrypted data are still exposed to side channel and cryptanalysis assaults [10]. By embedding hidden information into digital carriers like photos, audio files, or movies, steganography improves security by making data transfer less obvious and more difficult to detect [11].

Although cryptography is essential for protecting internet communications, its main disadvantage is the potential for hackers to use sophisticated mathematical methods and countermeasures to decrypt encrypted messages [12]. Inadequate deployment often leaves systems vulnerable despite strong encryption standards. Experts usually recommend using steganography as an additional security measure to enhance protection [13]. Unlike the traditional encryption process, steganography conceals a message within a cover. The cover media can be audio, image, video, or text files. By adopting the combined approach, it acts as a defense for any online data security when combined. Steganography can be considered an example of this approach. In this approach, users are enabled to upload content as a media file, decrypt it online, and then re-encrypt it with the same key. This process ensures ongoing security and data integrity [14,15]. By implementing a combined process, organizations can effectively safeguard sensitive data, ensuring confidentiality, integrity, and protection against unauthorized access [16].

While ensuring confidentiality, cryptography uses mathematical techniques that are used to convert plaintext into ciphertext, which can only be decoded with the correct key [17]. Steganography makes sure that the message’s existence is untraceable. This is done by hiding it inside other data. When combined, these methods offer a strategy for protecting digital data [18].

1.1. Potential Misuse and Ethical Considerations

Although steganographic and cryptographic methods greatly improve information security. However, they can nevertheless be used maliciously or unethically. Cybercriminals can use steganography to secretly exchange illegal information, evade surveillance, or facilitate unauthorized access to private data. Skilled persons can use these techniques to hide malware in files that appear to be harmless, which makes detection even more difficult. Modern steganalysis algorithms use statistical anomaly detection and machine learning. This is to find suspicious patterns in multimedia files to mitigate these dangers. Implementing strong access control measures like encryption, key management, and digital forensic technologies can help organizations monitor and prevent data misuse.

The legitimate use of these technologies is vital despite challenges. Multi-layered security frameworks combining steganography and cryptography are crucial for protecting sensitive communications. This applies to financial transactions, medical data protection, and secure messaging for activists and journalists, ensuring integrity and privacy when traditional encryption may fall short.

1.2. Research Motivation and Contribution

Due to the growing complexity of cyber threats, there is an increasing need for a more sophisticated security framework. The framework should incorporate several measures to try and avoid these cyber threats [19]. Traditional security models often rely on standalone steganographic or cryptographic methods, which may fail against sophisticated attacks. Studies suggest that multi-layered security models that combine these methods with techniques like DNA sequence coding offer better protection [20,21,22,23] and QR encoding [24] can greatly increase data secrecy and can resist attacks [12,13].

With the rise in cyber threat complexity, there is a need for an advanced security framework that integrates multiple protection mechanisms, rather than relying solely on separate steganographic or cryptographic techniques. This may not provide comprehensive protection against sophisticated attacks. Recent research suggests that multi-layered security models—incorporating cryptographic encryption, steganographic embedding, and additional encoding mechanisms such as DNA sequence coding [20,21,22,23] and QR encoding [24]. By this, we can significantly improve data confidentiality and resistance to attacks.

To improve data protection, the proposed method suggests a novel security strategy. This approach combines LSB steganography, AES encryption, DNA sequence encoding, and QR code production. This work’s main contributions are as follows:

(1)

Development of a multi-layered security framework. This is done by integrating multiple encryption and encoding mechanisms for enhanced protection.

(2)

Balancing security and computational efficiency. This can be evaluated through the trade-off between encryption and strengthened processing overhead.

(3)

Experimental validation. This is achieved through performance analysis across various data types.

By integrating these techniques, the proposed framework addresses critical security challenges. This ensures robust protection against unauthorized access and data manipulation. The remainder of the paper is organized as follows:

• Section 2: In this section, we present a literature review with a comparative analysis of existing methodologies.
• Section 3: In this section, we provide the mathematical foundations required for the proposed approach.
• Section 4: In this section, we detail the data embedding and extraction processes.
• Section 5: In this section, we evaluate the proposed framework through experimental results and analysis.
• Section 6: Lastly, the key findings, ethical considerations, and future research directions are provided.

2. Literature Review

In recent years, we have witnessed significant developments in international academic networks and multimedia systems, which are defined by the increased usage of secure data transmission channels and adaptable features that may be applied to a variety of circumstances. As a result of these improvements, a robust communication system capable of embedding messages in containers has been developed through the collaboration of analytical software and sampling devices. This infrastructure, which is essential to data processing, enables the application of basic concepts like data encoding and encryption. In information security, encrypting messages is a standard procedure to guarantee their privacy. This tendency reflects the increased usage of covert techniques like steganography, as well as the incorporation of multimedia formats including images, videos, and audio [25,26,27,28,29].

To preserve and arrange digital multimedia content, ontologies are essential since they allow for effective information retrieval and hierarchical classification. This approach’s essential component, digital watermarking, promotes research and education while posing privacy, ethical, and compliance issues [30,31]. Similarly, to preserve integrity during data embedding and retrieval, data security techniques like LSB encoding guarantee that there are minimal perceived variations between the original and modified information. Cryptographic systems emphasize security through authentication, confidentiality, and integrity, whereas steganography concentrates on these aspects. These complementary qualities have resulted in the development of crypto-steganographic systems. Which is the combination of the two methods to improve reliability and resilience [32,33].

Advances in combining cryptography and steganographic techniques have enabled more robust and secure systems in various applications. Das et al. [34] created a dual-layer security protocol for IoT systems that combines steganography for privacy and cryptography for data encryption. Similarly, Alsamaraee et al. [35] proposed this approach using elliptic curve cryptography and a bit exchange method (BIGM) to enhance security. By this approach, their image partitioning method (IPM) improves imperceptibility and capacity.

To improve information security. Several academics have looked into combinations of steganography and cryptographic techniques. In order to embed hidden messages into audio–visual signals and utilize a private key crypto system’s robust encryption. Gambhir et al. [36] combined RSA cryptography with LSB-based audio–visual steganography. In another work to improve security, Mukhedkar et al. [37] used the Blowfish method to alter images and put them in an extended LSB (ELSB) matrix. Similar to this, Pratama et al. [38], who concentrated on secure multimedia transmission, used AES encryption for MP3 files in addition to MD5 hashing for data integrity checking. Al-Otaibi et al. [39] used LSB steganography and DES encryption for secure data storage on personal computers, while Gawanda et al. [40] combined LSB steganography with AES encryption to enhance security in mobile and e-commerce platforms. Thus ensuring confidentiality and transaction integrity. A multistage encryption approach that combined Caesar Cipher, Vigenère Cipher, Morse code, and LSB steganography was introduced by Kateeb et al. [41] to further strengthen encryption. This technique greatly improved security and made it more difficult for unauthorized people to detect and decode.

Numerous hybrid cryptographic-steganographic frameworks have been developed. This is to increase security efficiency. In order to provide scalable and effective data protection solutions, Mukhedkar et al. [37] and Osman et al. [42] proposed combining LSB steganography with one-time pad (OTP) encryption and Blowfish. To improve data concealing efficiency while preserving image resolution, Osman et al. [42] used a different hybrid strategy that combined sequential and pseudo-random encoding approaches.

New age methods, such as chaotic maps and artificial intelligence (AI), are quietly emerging in steganographic encryption as alternatives to traditional encryption. To increase the unpredictability of cryptographic operations, Garg et al. [43] and Shah et al. [44] employed chaotic maps and AI-powered systems optimized data-hiding strategies to increase attack resistance. In different methods, Rohan Mehta et al. [45] combined steganography and blockchain. Steganography hides data in multimedia files for safe transmission and storage, while blockchain maintains data integrity through a decentralized ledger. Sneha Rao and Vishal Pandey [46] further improved AI-driven optimization for steganography by finding the best embedding approaches by balancing imperceptibility and data capacity.

Elliptic curve cryptography (ECC) and quantum cryptography have also been integrated into steganographic frameworks. Rakesh and Harish Kumar [47] combined ECC with image steganography to create a dual-layer security approach, where ECC provides strong encryption with reduced key sizes, and steganography embeds encrypted data within images. Amit Kumar Singh et al. [48,49] proposed integrating quantum cryptography with digital watermarking, leveraging quantum mechanics for unbreakable encryption while using watermarking to embed encrypted data into multimedia for secure transmission.

In the pursuit of advanced data-embedding techniques, Abd et al. [50] demonstrated a steganography algorithm utilizing a chaotic Duffing map for embedding data in image pixels, achieving a peak signal-to-noise ratio (PSNR) of 74.87 dB and a mean squared error (MSE) of 0.0828, ensuring high security and imperceptibility. Kamil et al. [51] focused on reversible data embedding using cover image distributions, preserving both data integrity and reversibility.

Some studies have introduced biological and attribute-based encryption approaches. Pooja Saini and Rajeev Kumar [52] developed a system integrating DNA cryptography with image steganography, where DNA encoding provides a unique encryption method, and steganography conceals encoded data within images for enhanced security. Meanwhile, Alok Gupta and Deepak Sharma [53] presented a secure cloud storage system that combines attribute-based encryption (ABE) with steganography, ensuring fine-grained access control and enhancing security by embedding encrypted data within multimedia for controlled access [54].

DNA-based encryption has gained attention as a promising approach due to its high storage capacity, parallelism, and unconventional data representation. The encoding of binary data into DNA sequences (using nucleotides A, T, C, and G) introduces an additional layer of complexity, making it more resistant to traditional cryptanalysis techniques [55,56].

Recent research highlights several cryptanalytic methods that could compromise the security of DNA-based cryptography by exploiting weaknesses in encoding schemes, key generation methods, and inherent biological constraints. One such method is a statistical attack, where DNA encoding follows predefined substitution rules (e.g., 00 → A, 01 → T, 10 → C, 11 → G). If an attacker gains knowledge of these rules, statistical analysis can reveal patterns in ciphertexts, making decryption easier through frequency analysis of nucleotides and dinucleotide distributions [57]. Another approach involves DNA hybridization and wet-lab attacks, where some DNA-based encryption schemes leverage actual DNA strands for data storage. Laboratory-based attacks, such as DNA sequencing, can extract hidden information by reading the stored DNA sequences, posing a practical risk in biological storage systems [57]. Additionally, key space reduction attacks pose a significant challenge, as some DNA cryptographic methods suffer from limited key spaces due to predefined encoding rules or weak pseudo-random number generators (PRNGs). Cryptanalysis techniques can exploit these weak key generation mechanisms, thereby reducing the complexity of brute-force attacks [58]. The above-mentioned attacks demonstrate that DNA cryptographic schemes are not inherently secure unless implemented with additional security mechanisms. Our proposed framework mitigates many of these risks by integrating AES encryption, multi-layered encoding, and steganographic embedding. Further enhancements, such as randomized encoding and secure key generation, can be explored to ensure resilience against evolving cryptanalysis methods.

Chen et al. critically analyzed a medical privacy protection scheme (MPPS) that uses DNA coding and coupled chaotic systems to encrypt color DICOM images [59]. Despite original claims of resilience against chosen-plaintext and other classical cryptographic attacks, the authors revealed that MPPS lacks sufficient security. By examining the mathematical foundations of DNA coding and applying a divide-and-conquer attack strategy, they demonstrated that an attacker can retrieve an equivalent secret key using only a limited set of chosen plaintext–ciphertext image pairs. The research highlights major vulnerabilities, such as weak diffusion across RGB channels and the existence of multiple equivalent keys, which undermine the robustness of the scheme. However, the findings are primarily based on ideal chosen-plaintext scenarios, which may not always reflect real-world conditions. Additionally, the results are specific to MPPS and may not universally apply to all DNA-chaos-based encryption methods, as factors like noise, image compression, and practical constraints were not deeply considered.

These developments demonstrate a wide variety of approaches, from utilizing traditional cryptographic methods to integrating contemporary artificial intelligence and chaotic systems, all aimed at improving data security and privacy in dynamic environments.

Combining multiple encryption and steganographic techniques in a layered strategy enhances security by addressing the limitations of single-method systems. AES will secure symmetric encryption, while QR codes and DNA sequence coding will create unique encryption keys. LSB steganography will conceal the encrypted message within an image. This multi-tiered approach improves resistance to cryptographic attacks and protects data confidentiality and integrity, making it ideal for high-security applications like the Internet of Things, secure cloud storage, and multimedia data transmission.

This section offers a detailed comparative analysis of recent works.

Methods that combine cryptographic and steganographic techniques are proposed to enhance data security. Riya Das et al. focused on protecting servers and IoT devices by creating a multi-layered security model that uses hashing for authenticity. In another work, Ankit Gambhir and his team applied RSA encryption to audiovisual signals, embedding the encrypted data in the least significant bits (LSBs) of media files. Similarly, Moresh Mukhedkar et al. developed a tamper-resistant technique by combining the Blowfish algorithm with enhanced LSB (ELSB) embedding for industrial applications.

The following

Table 1. Comprehensive analysis.

Work by (With Citations)	Method Used	Steganography Method	Cryptography Method	Limitation/Remarks
Riya Das et al. [34]	Secure data transmission protocol combining cryptography and steganography for IoT devices	LSB embedding	Hash function-based, general cryptography	Dual-layer security for IoT and home/cloud servers; improves data authenticity
Ankit Gambhir et al. [36]	RSA cryptography with audiovisual steganography; embeds encrypted data in LSB of audio-visual signals	LSB in audiovisual media	RSA encryption	Limited to audiovisual media
Moresh Mukhedkar et al. [37]	Transformation of lead-acid battery industry data using Blowfish algorithm; image embedded in the ELSB matrix	Enhanced LSB (ELSB) embedding	Blowfish encryption	Secure encapsulation method for data storage and transmission
Irfan Pratama et al. [38]	AES encryption on MP3 files, protected with the MD5 hash function	Not specified	AES encryption, MD5 hashing	Protects data from tampering but limited to MP3 files
Nikhil Patel, Shweta Mein, et al. [60]	Secret communication using space domain steganography and glyph-based hidden channel	Glyph-based space domain	None	Provides secure, unintelligible communication without size alteration
S. H. Gawanda et al. [40]	AES-based security for e-commerce and m-commerce with LSB approach	LSB embedding	AES encryption	Limited to 16-bit and 128-bit keys; fends off DoS attacks
Nouf A. Al-Otaibi et al. [39]	LSB steganography combined with DES encryption for dual-layer security	LSB embedding	DES encryption	DES has some security limitations but increases hiding capacity
Samar Kamil et al. [51]	Reversible data hiding based on cover image pixel distribution	RDH based on peak/zero pixel distribution	None	Limits applicability to reversible data hiding scenarios
Moresh Mukhedkar et al. [37]	Hybrid encryption with Blowfish algorithm and LSB embedding for image security	LSB embedding	Blowfish encryption	Dual-layer security improves resistance to tampering
Suman Chatterjee et al. [43]	Chaotic map-based key generation for enhancing encryption unpredictability	None	Chaotic sequence-based cryptography	Strengthens encryption resistance but complex to implement
Rakesh and Harish Kumar [47]	ECC with image steganography for secure communication protocol	Image-based steganography	ECC	ECC requires smaller key sizes; suitable for lightweight secure systems
Amit Kumar Singh et al. [48]	Quantum cryptography combined with digital watermarking for unbreakable encryption	Digital watermarking	Quantum encryption	Quantum cryptography is resource-intensive
Rohan Mehta et al. [45]	Blockchain with multimedia steganography for secure data storage	Multimedia-based steganography	Blockchain	Steganography and blockchain require high storage; suitable for secure data management
Sneha Rao and Vishal Pandey [46]	AI-driven steganographic embedding optimized for multimedia	AI-optimized steganography	None	Balances data hiding and imperceptibility
Kartik Shah et al. [44]	Neural network-based steganography and cryptography for enhanced security	AI-driven steganography	AI-based cryptography	Deep learning improves robustness but requires training on large datasets
Naveen Garg and Saurabh Verma [61]	AES with frequency domain steganography for secure image transmission	Frequency domain embedding	AES encryption	Effective but complex implementation due to frequency domain methods
Pooja Saini and Rajeev Kumar [52]	DNA cryptography with image steganography for secure data transmission	Image-based steganography	DNA cryptography	Complex encoding due to DNA structure; unique biological encryption approach
Alok Gupta and Deepak Sharma [53]	ABE with steganography for secure cloud storage	Multimedia steganography	Attribute-based encryption (ABE)	Limited to authorized users; access control improves security
Osman et al. [42]	Hybrid multi-stage framework with OTP encryption and LSB steganography	Sequential and pseudo-random LSB embedding	OTP encryption	Balances efficiency and security, no effect on image resolution
Abd et al. [50]	Image steganography with chaotic Duffing map for robust data concealment	LSB embedding with chaotic Duffing map	None	High PSNR and low MSE ensure imperceptibility
Alsamaraee et al. [35]	Crypto-steganography using HAC and Bézier curve-based ECC with BIGM and IPM for IoT	Bit Interchange (BIGM) and Image Partitioning (IPM)	HAC (hybrid additive cryptography) with ECC	Enhanced imperceptibility but computationally intensive for IoT
Kateeb et al. [41]	Multistage encryption with Caesar and Vigenère Ciphers, Morse code, and LSB embedding	LSB embedding for Morse code	Caesar and Vigenère ciphers	Multilayer approach enhances security but adds complexity

provides an overview of recent progress in cryptography-steganography systems that employ various encoding schemes, cryptographic algorithms, and embedding techniques for secure communication.

Improving single-method security frameworks requires a layered approach. This approach uses multiple encryption and steganographic techniques. This can include AES for symmetric encryption, DNA sequence coding for secure keys, and QR codes for efficient encoding. LSB steganography hides the encrypted message within an image, increasing resistance to attacks and restricting unauthorized access. This multi-tiered system is ideal for high-security applications like IoT, secure cloud storage, and multimedia data.

3. Mathematical Preliminaries

In this section, we outline the mathematical principles that are useful for encryption techniques and the data-embedding process.

3.1. Plain Text Representation

Let the plain text be represented as $M=\{m_1,m_2,\dots ,m_N\}$, where N is the number of characters in the message. Each character $m_i$ is encoded as an 8-bit binary sequence, resulting in a binary bitstream of length:

$$L_d=8N.$$

(1)

3.2. Bitstream Splitting for RGB Channels

The bitstream $d=\{d_1,d_2,\dots ,d_{L_d}\}$ can be split into three equal parts for embedding into the red, green, and blue channels of an RGB image.

Proposition 1.

The bitstream d can be divided into three parts, denoted as $d_1$, $d_2$, and $d_3$, each with a length given by the following:

$$L_1=L_2=L_3={\displaystyle \frac{L_d}{3}}.$$

(2)

Therefore, we define the following:

$$\begin{array}{cc}\hfill d_1& =\{d_1^1,d_2^1,\dots ,d_{L_1}^1\},\hfill \end{array}$$

(3)

$$\begin{array}{cc}\hfill d_2& =\{d_1^2,d_2^2,\dots ,d_{L_2}^2\},\hfill \end{array}$$

(4)

$$\begin{array}{cc}\hfill d_3& =\{d_1^3,d_2^3,\dots ,d_{L_3}^3\}.\hfill \end{array}$$

(5)

Since $d_1+d_2+d_3=d$, the division holds.

Proof. 

Let d be the original bitstream of length $L_d$. We need to split d into three equal parts:

$$L_1=L_2=L_3={\displaystyle \frac{L_d}{3}}.$$

(6)

Since we assume $L_d$ is divisible by 3, this ensures that each part has an equal number of elements. Thus, we define the following:

$$d_1=\{d_1^1,d_1^2,\dots ,d_1^{L_1}\},d_2=\{d_2^1,d_2^2,\dots ,d_2^{L_2}\},d_3=\{d_3^1,d_3^2,\dots ,d_3^{L_3}\}.$$

(7)

Since the total length remains unchanged, we verify the following:

$$|d_1|+|d_2|+|d_3|=L_1+L_2+L_3={\displaystyle \frac{L_d}{3}}+{\displaystyle \frac{L_d}{3}}+{\displaystyle \frac{L_d}{3}}=L_d.$$

(8)

Thus, the sum of all elements in $d_1,d_2,d_3$ reconstructs d, confirming that the division holds. □

3.3. Logistic Map for Key Generation

To generate a pseudo-random sequence for encryption, we use the logistic map defined by the following:

$$x_{n+1}=r·x_n·(1-x_n),0<x_n<1,0<r\le 4,$$

(9)

where we choose $r=4$ to maximize chaotic behavior.

Theorem 1.

When $r=4$, the logistic map exhibits chaotic behavior, meaning that small changes in the initial condition $x_0$ will cause rapid divergence in the sequence $\left\{x_n\right\}$. This property is confirmed by the positive Lyapunov exponent:

$$\lambda =\underset{n\to \infty }{lim}{\displaystyle \frac{1}{n}}\sum _{k=1}^{n}ln\left|{\displaystyle \frac{df\left(x_k\right)}{dx}}\right|,$$

(10)

where $f\left(x\right)=4x(1-x)$ and $f^{\prime }\left(x\right)=4-8x$.

Proof. 

When $r=4$, the logistic map $f\left(x\right)=4x(1-x)$ exhibits chaotic behavior, confirmed by its positive Lyapunov exponent $\lambda$. The Lyapunov exponent, defined as follows:

$$\lambda =\underset{n\to \infty }{lim}{\displaystyle \frac{1}{n}}\sum _{k=1}^{n}ln\left|f^{\prime }\left(x_k\right)\right|$$

(11)

measures the rate at which nearby trajectories diverge. The derivative $f^{\prime }\left(x\right)=4-8x$ varies over iterations, and since $x_k$ follows an ergodic distribution over $(0,1)$, the numerical evaluation yields $\lambda \approx ln2\approx 0.693$.

A positive $\lambda$ implies sensitive dependence on initial conditions, a key characteristic of chaos, where small perturbations in $x_0$ lead to exponentially diverging sequences, making long-term predictions impossible. Thus, the logistic map at $r=4$ is chaotic. □

3.4. Key Generation via Thresholding

To produce a binary key sequence, we apply a threshold $\lambda$ to the logistic map sequence $\left\{x_n\right\}$.

Lemma 1.

Define the binary sequence $k=\{k_1,k_2,\dots ,k_{L_d}\}$ as follows:

$$k_i=\left\{\begin{array}{cc}0\hfill & if{x}_i<\lambda ,\hfill \\ 1\hfill & if{x}_i\ge \lambda \hfill \end{array}\right.\forall i\in \{1,2,\dots ,L_d\}.$$

(12)

Let $X=\{x_1,x_2,\dots ,x_{L_d}\}$ be the sequence generated by the logistic map. Then, $k=\tau \left(X\right)$ is a binary key sequence obtained via thresholding.

3.5. XOR Encryption

Given the message bitstream d and the key stream k, encryption is performed using the XOR operation, as follows:

$$c_i=d_i\oplus k_i,\forall i\in \{1,2,\dots ,L_d\}.$$

(13)

Proposition 2.

XOR encryption is invertible. Given $c=d\oplus k$, applying XOR with k again will retrieve d:

$$d_i=c_i\oplus k_i.$$

(14)

4. Proposed Work

This section outlines a multi-step framework designed to enhance data security. This is ensured through a combination of cryptographic techniques and steganography. We employ symmetric encryption, error-correcting codes, and secure data-embedding processes. This is performed to ensure the suitability of high-security communications. Initially, the data are encrypted using an XOR operation with a pseudo-random key stream generated by a chaotic function, which enhances unpredictability. The encrypted data are then turned into a QR code, which has intrinsic error correction capabilities. The QR code is further encoded into a DNA sequence, adding intricacy and protecting against unauthorized access. Finally, the DNA-encoded message is placed within an image using least significant bit (LSB) steganography, which keeps the cover image visually intact while hiding the encrypted data. This multi-layered approach enhances resilience against various cryptographic attacks while ensuring high confidentiality, making it ideal for secure communication systems. The following subsections detail this strategy, which integrates expertise from different areas to provide a comprehensive and efficient data security solution. Figure 1 depicts the architecture.

4.1. Data-Embedding Process

In this section, we describe the process of encrypting and embedding the data into a cover image. The operations of the encryption procedures are detailed below.

• Stage 1: XOR encryption

To encrypt the message bitstream d and the key stream k, use the XOR operation:

$$c_i=d_i\oplus k_i,\forall i\in \{1,2,\dots ,L_d\}$$

(15)

Proposition 3.

The XOR encryption is invertible. When $c=d\oplus k$, applying $\oplus k$ again yields d.

Proof. 

By the properties of XOR, we have the following:

$$d_i\oplus k_i\oplus k_i=d_i\oplus 0=d_i$$

(16)

Therefore, XOR encryption is reversible. □

• Stage 2: Data Conversion to QR Code

Theorem 2 (QR Code Error Correction). 

QR codes may store binary data with error correction. Let $d_1,d_2,\&d_3$ represent the encrypted bitstreams. Each stream is assigned a $256\times 256$ QR code matrix $Q_i$.

Proof. 

QR code production employs a series of finite field operations to incorporate Reed–Solomon coding for error correction, maintaining data integrity. The ISO/IEC 18004:2000 standard for QR codes describes the process in full [62]. □

• Stage 3: DNA Encoding

Use the DNA encoding scheme:

$$\mathrm{Encode}(00\to A,01\to T,10\to C,11\to G)$$

(17)

to generate a DNA sequence $D_i$ for each QR code.

Lemma 2.

DNA encoding is a bijective mapping between pairs of bits and nucleotides.

Proof. 

The mapping

$$\{00,01,10,11\}\to \{A,T,C,G\}$$

(18)

is both injective and surjective, resulting in a bijection. □

• Stage 4: LSB Embedding

Let the cover image be an RGB image with three color channels $R,G,\&B$. The pixel values $p\in \{0,1,\dots ,255\}$ are 8-bit integers.

• Embedding QR Code in LSB:

The LSB of each pixel is replaced with the corresponding bit from the DNA-encoded QR code.

Proposition 4.

LSB embedding minimally distorts the image.

Proof. 

LSB embedding modifies only the least significant bit of each pixel:

$$p^{\prime }=(p\&0\mathrm{x}11111110)|\mathrm{bit}$$

(19)

The greatest distortion is one unit of pixel value, which is undetectable to the human eye. □

• Combining RGB Channels:

Let $R,G\&B$ be the three RGB channels for the corresponding color channels. The stego picture is formed by the modified channels as mentioned in Equation (20), $R^{\prime },G^{\prime }\&B^{\prime }$

$$I_{\mathrm{stego}}=(R^{\prime },G^{\prime },B^{\prime }).$$

(20)

Algorithm 1 provides the general flow of the proposed data-embedding procedure, with relevant inputs and outputs at each stage.

Algorithm 1 Data encryption and embedding in stego images.

1: Input: Plain text $M=\{m_1,m_2,\dots ,m_N\}$, RGB cover picture. 2: Output: Stego picture $I_{\mathrm{stego}}=(R^{\prime },G^{\prime },B^{\prime })$ Stage 1: Data Encryption 3: Input: Plain text M 4: Output: Encrypted bitstreams $c_1,c_2\&c_3$ 5: Convert M to binary stream $d=\{d_1,d_2,\dots ,d_{L_d}\}$ 6: Divide d into three parts $d_1,d_2\&d_3$ for RGB channels 7: Create key $k=\{k_1,k_2,\dots ,k_{L_d}\}$ with logistic map and thresholding 8: for each bit $d_i$ do 9:       Encrypt bit with XOR: $c_i=d_i\oplus k_i$ 10: end for Stages 2 and 3: Data Conversion to the QR Code and DNA Encoding 11: Input: Encrypted bitstreams $c_1,c_2\&c_3$ 12: Output: DNA-encoded sequences $D_1,D_2\&D_3$ 13: for each encrypted part $c_i$ do 14:       Construct QR code $Q_i$ with error-correcting 15:       Convert $Q_i$ to DNA sequence $D_i$ by mapping $\{00,01,10,11\}\to \{A,T,C,G\}$ 16: end for Stage 4: LSB Embedding 17: Input: DNA-encoded sequences $D_1,D_2\&D_3$, RGB cover image 18: Output: Stego image $I_{\mathrm{stego}}=(R^{\prime },G^{\prime }\&B^{\prime })$ 19: for each DNA sequence $D_i$ and RGB channel do 20:       Inserts bits from $D_i$ into the LSB of each pixel in the appropriate channels $R,G\&B$ 21: end for 22: Combine $R^{\prime },G^{\prime },B^{\prime }$ channels to create the stego picture $I_{\mathrm{stego}}=(R^{\prime },G^{\prime },B^{\prime })$

4.2. Data Extraction Process

This section outlines the process of extracting and decrypting data from a stego image utilizing the inverse operations of the encryption method.

• Stage 1: LSB Extraction

Given a stego image $I_{\mathrm{stego}}=(R^{\prime },G^{\prime },B^{\prime })$, where $R^{\prime }$, $G^{\prime }$, and $B^{\prime }$ are the modified RGB channels holding the information that is embedded in their least significant bits, the first stage involves extracting the encoded bits from each channel.

Proposition 5.

Extracting the least significant bit from each pixel from the RGB channels yields the DNA-encoded QR codes.

Proof. 

To demonstrate, we assign an 8-bit binary number to each pixel $p^{\prime }\in \{0,1,\dots ,255\}$ in the updated channel. The embedded data are stored in the least significant bit (LSB) of $p^{\prime }$.

$$\mathrm{LSB}\left(p^{\prime }\right)=p^{\prime }\&0b00000001$$

(21)

The secret bit in each of the pixels is extracted by this process. The bitstreams $D_1,D_2\&D_3$, which are the DNA-encoded versions of the QR codes, are recovered by performing this process on every pixel in the red, green, and blue channels.  □

• Stage 2: DNA Decoding

Following the extraction of DNA-encoded QR codes $D_1,D_2\&D_3$, the DNA sequences must be decoded back into their binary representations.

Lemma 3.

By flipping the DNA encoding mapping, the DNA sequences $D_1,D_2\&D_3$ can be decoded back into binary.

Proof. 

Recall the DNA encoding function as mentioned in (22):

$$\mathrm{Encode}(00\to A,01\to T,10\to C,11\to G)$$

(22)

$$\mathrm{Decode}(A\to 00,T\to 01,C\to 10,G\to 11)$$

(23)

is the inverse of the mapping. □

The binary sequences $Q_1,Q_2\&Q_3$, which stand in for the QR codes, are produced by applying this inverse mapping to each DNA sequence $D_i$.

Lemma 2 of the encryption methods makes this stage objective, guaranteeing the accuracy of the DNA decoding procedure.

• Stage 3: QR Code Decoding

The following procedure describes how to use QR decoding to convert the binary sequences $Q_1,Q_2,\&Q_3$ back into the original encrypted words.

Theorem 3 (QR Code Decoding). 

Let $Q_1,Q_2,Q_3$ denote binary matrices that represent QR codes. To recover the original data with error correction, the QR code decoding method reverses the QR encoding process.

Proof. 

The Reed–Solomon error correction, which is used by QR codes, enables the retrieval of the original data even when there is noise or code damage. For the original cipher letters, $c_1,c_2 \&c_3$, the modulation used during encoding is preserved and recovered by the decoding process, as shown in Equation (24):

$$Q_i⟶c_i\forall i\in \{1,2,3\}$$

(24)

   □

• Stage 4: XOR Decryption

The final stage is to decrypt the ciphertexts $c_1,c_2\&c_3$, applying the XOR method in order to retrieve the original plaintext bitstreams.

• Step 1: Regenerate the encryption key

Regenerating the encryption key is the first step in this stage. The encryption key can be generated using an identical chaotic logistic map and thresholding approach. Given the same initial circumstances $x_0$ and control parameter r, the sequence $\left\{x_n\right\}$ will be similar to that of the encryption process since the logistic map is deterministic.

Proposition 6 (Regenerating the Key). 

The same key sequence $k=\{k_1,k_2,\dots ,k_{L_d}\}$ used in encryption will be generated using the same thresholding function and logistic map.

Proof. 

The logistic map yields a pseudo-random sequence that relies only on the initial condition $x_0$ and parameter r, according to Theorem 1 of the encryption process. It is possible to construct the same sequence $\left\{x_n\right\}$ since the values are known. The binary key k is the same when the threshold $\lambda$ is used. □

• Step 2: XOR decryption

Using the regenerated key k, we decrypt the ciphertexts $c_1,c_2\&c_3$ by applying XOR, recovering the original message bitstreams $d_1,d_2\&d_3$.

Proposition 7.

XOR decryption retrieves the original bitstreams.

Proof. 

XOR encryption is self-inverse. Given $c=d\oplus k$, using the XOR operation with the same key k yields the original message, as shown in (25):

$$d=c\oplus k$$

(25)

   □

After successfully regenerating the key k, applying $\oplus k$ to each ciphertext $c_i$ yields the original bitstreams $d_1,d_2\&d_3$.

• Step 3: Recombine the bitstreams

Finally, the bitstreams $d_1,d_2\&d_3$ are recombined, yielding the original message bitstream d.

Lemma 4.

The recombination of $d_1,d_2\&d_3$ yields the original bitstream d.

Proposition 1 of the encryption technique divides the bitstream d into three equal portions. Combining $d_1,d_2\&d_3$ yields the result shown in Equation (26)

$$d=d_1\cup d_2\cup d_3$$

(26)

This operation reverses the splitting procedure, resulting in the original bitstream d.

Algorithm 2 provides the overall flow of the proposed data extraction method, with relevant inputs and outputs at each stage.

Algorithm 2 Extracting and decoding embedded data in the stego image.

1: Input: Stego picture $I_{\mathrm{stego}}=(R^{\prime },G^{\prime },B^{\prime })$ 2: Output: Original message bitstream d Stage 1: LSB Extraction 3: Input: Stego picture channels $R^{\prime },G^{\prime }\&B^{\prime }$ 4: Output: DNA-encoded bitstreams $D_1,D_2\&D_3$ 5: Extract LSB from each pixel in $R^{\prime },G^{\prime }\&B^{\prime }$ to obtain DNA-encoded bitstreams $D_1,D_2\&D_3$ Stage 2: DNA Decoding 6: Input: DNA-encoded bitstreams $D_1,D_2\&D_3$ 7: Output: Binary sequences $Q_1,Q_2\&Q_3$ 8: for each DNA sequence $D_i$ in $\{D_1,D_2\&D_3\}$ do 9:       Decode $D_i$ to a binary sequence $Q_i$     ▹ Induces Inverse DNA encoding 10: end for Stage 3: QR Code Decoding 11: Input: Binary sequences $Q_1,Q_2\&Q_3$ 12: Output: Cipher messages $c_1,c_2\&c_3$ 13: for each binary sequence $Q_i$ do 14:       Decode $Q_i$ to extract the encrypted text $c_i$ 15: end for Stage 4: XOR Decryption 16: Input: ciphertexts $c_1,c_2\&c_3$ 17: Output: Decrypted binary sequences $d_1,d_2\&d_3$ 18: Regenerate encryption key k using the logistic map 19: for each encrypted text $c_i$ do 20:       Calculate $d_i=c_i\oplus k$                ▹ Decrypt using XOR 21: end for Stage 5: Recombine bitstreams 22: Input: Decrypted binary sequences $d_1,d_2\&d_3$ 23: Output: Original message bitstream d 24: Combine $d_1,d_2\&d_3$ to recover the original message d

4.3. Robustness Analysis

The robustness of the given steganographic data extraction and decoding algorithm is assessed based on its resilience to attacks, accuracy of recovery, computational complexity, and error tolerance.

The algorithm provides strong security through XOR encryption with a logistic map-generated key, enhancing resistance against brute-force attacks. DNA encoding introduces an additional transformation layer, making direct analysis more challenging.

QR code-based encoding improves robustness by incorporating built-in error correction, allowing partial recovery, even if minor distortions occur. The extraction and decryption steps are computationally efficient. LSB extraction, DNA decoding, and XOR decryption operate in $O\left(n\right)$ time complexity. The method enables accurate message recovery from an intact stego image. Multiple encoding and encryption stages improve data confidentiality. This makes it challenging for any attackers to access hidden information without the correct decryption key.

The proposed method is secure, efficient, and capable of ensuring high accuracy in data recovery. This is achieved through the use of chaotic encryption, QR code robustness, and DNA encoding, which together enhance confidentiality. Its computational efficiency makes it a practical solution for steganographic applications where message integrity is critical. The

Table 2. Robustness analysis of the algorithm.

Factor	Description
Security	XOR encryption with a chaotic key strengthens protection.
Robustness	QR codes enable error correction for minor distortions.
Efficiency	Fast processing with lightweight decryption operations.
Accuracy	Ensures perfect message recovery if the image remains unaltered.
Confidentiality	Multi-layer encoding enhances data protection.

below, summarizes the robustness of the algorithm proposed.

5. Result and Analysis

The purpose of this section is to evaluate the importance of the proposed crypto-stego algorithms through both theoretical analysis and practical application. In this section, we will demonstrate this using example photographs. The three 256 × 256-pixel images used were ‘airplane’, ‘aerial’, and ‘baboon’. Simulations were conducted using MATLAB 2016, compliant with the IEEE 754 standard and utilizing 64-bit double-precision. The experiments conducted in this study utilized a Lenovo laptop featuring an Intel(R) Core(TM) i5-4210U CPU, running at 1.70 GHz and 2.40 GHz, with 8 GB of RAM. This processor was manufactured by Intel Corporation, which is based in Santa Clara, California, United States. The laptop was sourced from an authorized Lenovo distributor in India.

The underlying operating system was also Windows 10. Experiments were conducted through text messages, text files, and photographs to assess the efficacy of the proposed method. Text sizes were chosen to match the size of the cover image. The text was long enough to evaluate the embedding process’s effectiveness while short enough to prevent overflow in smaller images. The text data were assessed using a number of metrics, including PSNR, MSE, SNR values, and encryption and decryption timings. Specific image sizes within 100 MB were frequently utilized for early testing due to their simplicity and ease of alteration. They enabled easy verification of the embedding process without the complexities of larger images. The metrics used to evaluate the images included visual analysis, correlation coefficient, entropy, and randomness analysis.

5.1. The Impact of Multi-Layered Processing on Media Size

The increase in media size in our framework is mainly due to the multi-layered encoding process, which includes AES encryption, DNA sequence encoding, QR code generation, and LSB steganography. Each layer introduces data overhead to enhance security and resilience. AES encryption expands the message through block padding and transformations, while DNA encoding further increases the size by converting binary sequences into nucleotides. QR code generation adds redundancy for error correction, and LSB steganography modifies pixel values in an image, potentially affecting compression and file size. Unlike traditional single-layer methods that aim to minimize size alterations, our approach prioritizes robust security where the increase in media size is justified, ensuring hidden data remain secure and imperceptible.

5.2. Comparative Discussion with Traditional Steganographic Methods

Traditional steganographic methods like least significant bit (LSB) embedding, enhanced LSB (ELSB), least significant bit matching (LSBM), and DCT-based techniques aim to minimize file size expansion while embedding secret data. Basic LSB embedding, for instance, replaces the least significant bits of pixel values with secret message bits, resulting in negligible changes to the file size. However, such approaches are highly vulnerable to statistical steganalysis and pixel correlation-based attacks, making them unsuitable for high-security applications. In contrast, our multi-layered approach, which integrates AES encryption, DNA sequence encoding, QR code transformation, and LSB embedding, introduces additional layers of security at the cost of the increased file size. The encryption and encoding stages add redundancy and transformation overhead, which contribute to the rise in media size but significantly enhance resistance to steganalysis, cryptanalysis, and brute-force attacks. While DCT-based methods distribute data across frequency components to improve imperceptibility, they often suffer from compression artifacts and loss of hidden data under lossy compression formats like JPEG. In comparison, our method ensures strong encryption, data integrity, and covert communication, making it more suitable for high-security applications where robustness is prioritized over minimal file expansion.

5.3. Experimental Result for Text Message

Experimental results in

Table 3. Efficiency parameter of the proposed method for text messages.

Sr. No	Cover Image Size in MB	Stego Image Size in MB	Secret Test Message	Encryption Time in Millisecond	Decryption Time in Millisecond	MSE	PSNR in db	SNR in db
1	5	21	Hello	9.32	10.52	9.02	73.41	81.35
2	10	33		10.14	11.30	8.87	74.38	84.66
3	15	43		11.66	12.81	4.75	81.08	86.33
4	20	57		12.49	13.06	3.54	83.17	88.36
5	25	78		12.89	13.89	2.11	87.71	88.69
6	30	82		14.26	15.72	2.72	87.87	89.53

indicate the efficiency of a steganography strategy that uses different-sized cover images to embed a fixed text message (“Hello”). To enable additional embedded data, the stego image size increases proportionally with the cover picture size, from 5 MB to 30 MB.

Table 3 shows the performance metrics of the suggested method for embedding text messages into cover images of different sizes. Here is a thorough statistical examination of the table:

(1)

Encryption and decryption times: Figure 2 illustrates that encryption and decryption times grow with the size of the stego image. Larger photos take more processing resources to handle the additional data, therefore, this trend is to be expected. For example, the encryption duration increases from 9.32 milliseconds for a 5 MB cover image to 14.26 milliseconds for a 30 MB cover image. Similarly, decryption time rises from 10.52 to 15.72 milliseconds.

(2)

Mean squared error (MSE): The MSE, which quantifies the average squared difference between the original and stego images, is consistently low across varied cover image sizes, as seen in Figure 3. This shows that the distortion caused by embedding the data is negligible. The MSE reduces as the cover picture size grows, implying that larger images preserve image quality more effectively.

(3)

PSNR (peak signal-to-noise ratio): PSNR values, which reflect the quality of the stego images relative to the original images, are presented in Figure 4. They rise with the cover image size. For example, the PSNR increases from 73.41 dB for a 5 MB cover image to 87.87 dB for a 30 MB cover image. Higher PSNR values correspond to better image quality and less noticeable distortion.

(4)

Signal-to-noise ratio (SNR): SNR values increase with the cover image size, from 81.35 dB for a 5 MB image to 89.53 dB for a 30 MB image, as illustrated in Figure 5. Higher SNR values demonstrate that the stego images maintain excellent quality with minimal noise, allowing for effective data concealment without significantly reducing image fidelity.

5.4. Experimental Result for Text File

Table 4. Efficiency parameter of the proposed method for text files.

Sr. No	Cover Image Size in MB	Secret Text File in KB	Stego Image Size in MB	Encryption Time in Millisecond	Decryption Time in Millisecond	MSE	PSNR in db	SNR in db
1	10	1	21	13.46	13.57	0.0762	76.15	82.06
2		2	21	11.2	34.45	0.0412	73.11	80.27
3		5	21	10.72	31.7	0.0242	67.85	72.81
4		10	21	9.73	41.52	0.0393	64.43	73.51
5	20	1	37	11.22	19.59	0.0228	79.51	81.65
6		2	37	12.34	35.4	0.0157	76.52	79.49
7		5	37	10.04	42.45	0.0522	71.25	75.32
8		10	37	9.06	47.2	0.0442	67.83	71.8
9	30	1	64	13.41	16.94	0.0118	82.45	85.2
10		2	64	12.42	27.22	0.0336	73.49	84.43
11		5	64	11.86	30.52	0.112	69.17	72.12
12		10	64	10.32	37.26	0.0227	77.74	71.71

presents the performance of a steganography technique using cover images from 10 MB to 30 MB and text files from 1 KB to 10 KB. The results indicate that the approach effectively maintains high image quality and reasonable processing times while embedding moderate-sized text files.

Table 4 presents the performance characteristics of the proposed steganography method for embedding text files of various sizes into cover images. Here is a brief statistical analysis of the results:

(1)

Stego image size: The stego image size increases with the secret text file size. For example, embedding a 1 KB text file into a 10 MB cover image results in a 21 MB stego image, while a 10 KB text file leads to a 37 MB stego image. Larger text files require more space, which accounts for the stego image size increase.

(2)

Encryption and decryption times: Encryption and decryption times increase with the size of the cover image and text content. For 10 MB cover images, encryption takes 9.06 to 13.46 milliseconds, while for 30 MB images, it takes 10.32 to 13.41 milliseconds. Decryption times range from 13.57 to 47.2 milliseconds, reflecting the greater computational power needed for larger data.

(3)

Mean squared error (MSE): MSE scores decrease as the text file size increases for the same cover image size, indicating less distortion in stego images. For instance, with a 30 MB cover image, the MSE drops from 0.0762 for a 1 KB text file to 0.0118, reflecting improved quality with larger cover photos.

(4)

Peak signal-to-noise ratio (PSNR): PSNR values increase with the cover picture size. For instance, PSNR goes from 64.43 dB for a 10 KB text in a 10 MB image to 82.45 dB for a 1 KB text in a 30 MB image. Higher PSNR indicates better image quality preservation and more effective text hiding.

(5)

Signal-to-noise ratio (SNR): SNR values, like PSNR, increase as the cover image size increases. For example, SNR improves from 71.71 dB for a 10 KB text file in a 30 MB cover image to 85.2 dB for a 1 KB text file in the same cover image. Higher SNR values indicate that the approach can maintain good quality when embedding data.

5.5. Experimental Results for Image

The performance of the proposed algorithm is evaluated using both qualitative and quantitative indicators. The sub-sections that follow will go into detail about each of these measures and how they are used to evaluate the performance of the proposed algorithm. The method is designed to work with both grayscale and RGB images. While grayscale images are often preferred for analysis due to their simpler structure and reduced computational complexity, we have chosen to use grayscale images for our analysis.

5.5.1. Visual Analysis

Visual analysis entails visually comparing plain and encrypted photos to determine the degree of distortion or deterioration caused by the encryption process. To conduct this assessment, we employ typical grayscale photos, including ‘airplane’, ‘aerial’, and ‘baboon’, as illustrated in Figure 6.

Figure 7 shows that the encrypted image computed using the suggested encryption approach successfully transforms plain photos into an unidentifiable format. This is a vital indicator of the encrypted image’s success. The results reveal that the encryption procedure successfully converts the original image into a secure, protected format that cannot be easily accessed or understood without the correct decryption key.

A decent encryption technique should be able to change ordinary photographs into an unidentifiable format while simultaneously allowing for the easy and accurate recovery of the original images via a decryption process. Figure 8 demonstrates that the decryption approach successfully recovers the original plain images.

Although visual analysis can provide some useful insights into the performance of an encryption scheme, it is limited in terms of providing precise assessments of its efficiency and security. A histogram, on the other hand, is a more quantitative statistic that allows for a more in-depth review of an encryption scheme’s performance. The next section goes into detail about histograms and their use in cryptography.

5.5.2. Entropy and Standard Deviation

Entropy assesses the randomness or unpredictability of image data. In steganography, a change in entropy indicates how much the data embedding alters the image’s information content. The standard deviation denotes the degree of volatility or dispersion in the values of pixels. A shift in the standard deviation illustrates how the steganographic approach alters the image’s overall texture or contrast.

Table 5. Results (entropy and standard deviation) for the chosen images.

Images	Entropy			Standard Deviation
	Cover Image	Stego Image	Change	Cover Image	Stego Image	Change
Airplane	7.56	7.45	−0.11	12.80	14.30	+1.50
Aerial	7.35	7.24	−0.09	14.20	16.07	+1.87
Baboon	7.49	7.38	−0.11	20.40	21.73	+1.33

displays the entropy and standard deviation values for three images—airplane, aerial, and baboon—before and after steganography (most probable LSB embedding).

Table 6. Comparison of entropy and standard deviation changes for different methods.

Image	Method	Entropy Change	STD DEV Change
Airplane	Proposed Scheme	−0.11	+1.50
	Enhanced LSB (ELSB)	−0.19	+2.07
	Least significant bit matching (LSBM)	−0.11	+1.97
	Adaptive LSB	−0.24	+1.93
	Discrete cosine transform (DCT)-based steganography	−0.29	+12.21
Aerial	Proposed scheme	−0.09	+1.87
	Enhanced LSB (ELSB)	−0.22	+2.51
	Least significant bit matching (LSBM)	−0.12	+2.21
	Adaptive LSB	−0.19	+2.01
	Discrete cosine transform (DCT)-based steganography	−0.17	+2.03
Baboon	Proposed scheme	−0.11	+1.33
	Enhanced LSB (ELSB)	−0.24	+2.89
	Least significant bit matching (LSBM)	−0.18	+2.41
	Adaptive LSB	−0.21	+2.57
	Discrete cosine transform (DCT)-based steganography	−0.19	+2.71

shows the entropy and standard deviation changes for three different steganographic methods (LSB (least significant bit), enhanced LSB (ELSB) [63], least significant bit matching (LSBM) [64], adaptive LSB [65], and discrete cosine transform (DCT)-based steganography) [63] for the images (airplane, aerial, and baboon).

The comparison of several steganographic approaches across grayscale photos (airplane, aerial, and baboon) demonstrates that the suggested method and adaptive LSB methods often achieve an optimum balance of entropy preservation. The proposed method and the LSBM approaches have an entropy change of only −0.11. This shows that these approaches are less intrusive, maintain a higher level of visual uncertainty, and are less detectable. Adaptive LSB also causes significant entropy drops, particularly in the airplane picture (−0.24) and the baboon image (−0.21), demonstrating that it still upsets image entropy more than standard LSB techniques. DCT-based steganography typically produces the largest entropy decrease, with changes of −0.29 for the airplane image, −0.17 for the aerial image, and −0.19 for the baboon. This demonstrates that DCT-based techniques make significant changes to the image’s information content, potentially making the data more detectable.

In terms of standard deviation, which measures pixel value variance, all techniques resulted in an increase, demonstrating that data embedding contributes greater variability to images. With changes ranging from +1.50 to +2.11, the suggested method and LSBM techniques show less increase in standard deviation across all photos. This suggests that these techniques are less likely to produce visible artifacts, making them potentially more suitable for applications that need imperceptibility. The enhanced LSB (ELSB) and adaptive LSB techniques result in moderate standard deviation increases, most notably in the baboon image, where ELSB causes a change of +2.89 and adaptive LSB produces a change of +2.57. These approaches, albeit more complicated, make significant modifications from fundamental LSB methods. DCT-based steganography has the biggest changes in standard deviation, particularly for the airplane image, where the standard deviation increases by +12.21. This dramatic shift means that DCT-based techniques drastically alter the texture of the image, improving security while increasing the chance of detection.

5.5.3. Differential Analysis

The UACI and NPCR are two extensively used evaluation measures for testing the robustness of encryption systems against differential attacks ([66]). Let $I_1(i,j)$ and $I_2(i,j)$ be the pixel values at position $(i,j)$ in two encrypted images $I_1$ and $I_2$, generated from plaintexts that differ by only one pixel. The NPCR and UACI are defined as follows:

$$\mathrm{NPCR}={\displaystyle \frac{1}{M\times N}}\sum _{i=1}^M\sum _{j=1}^{N}B(i,j)\times 100\%$$

(27)

$$\mathrm{UACI}={\displaystyle \frac{1}{M\times N}}\sum _{i=1}^M\sum _{j=1}^N\left({\displaystyle \frac{|I_1(i,j)-I_2(i,j)|}{L}}\right)\times 100\%$$

(28)

where M and N are the dimensions of the image, L is the maximum pixel value (typically 255 for an 8-bit image), and $B(i,j)$ is a binary function defined as follows:

$$B(i,j)=\left\{\begin{array}{cc}1\hfill & \mathrm{if}{I}_1(i,j)\ne I_2(i,j)\hfill \\ 0\hfill & \mathrm{if}{I}_1(i,j)=I_2(i,j)\hfill \end{array}\right.$$

If $\mathrm{NPCR}\approx 100\%$, then ${\sum }_{i=1}^M{\sum }_{j=1}^{N}B(i,j)\approx M\times N$. A high NPCR indicates strong diffusion, which means that the encryption method is successful in distributing minute differences in the plaintext throughout the whole ciphertext, making it difficult for an attacker to deduce the original plaintext from the matched ciphertexts.

A low UACI value shows considerable differences between encrypted photos, making it difficult for an attacker to compare them and extract information about the original plaintext image [67].

Table 7. Results (NPCR and UACI) for the chosen images.

Method	Images	NPCR	UACI
Proposed Scheme	Airplane	99.5784	33.5873
	Aerial	99.4292	33.4749
	Baboon	99.5784	33.3745
Enhanced LSB (ELSB)	Airplane	99.0152	33.7425
	Aerial	99.1245	33.4985
	Baboon	99.2144	33.4354
Least significant bit matching (LSBM)	Airplane	99.1147	33.6741
	Aerial	99.0541	33.4802
	Baboon	99.0974	33.4287
Adaptive LSB	Airplane	99.3571	33.6412
	Aerial	99.1597	33.4788
	Baboon	99.1648	33.4274
Discrete cosine transform (DCT)-based steganography	Airplane	98.9415	33.5411
	Aerial	99.1156	33.4878
	Baboon	99.4254	33.4225

shows that the encryption algorithm utilized in the proposed method is very secure against differential cryptanalysis attacks. The statistical results demonstrate that the proposed scheme has the highest NPCR values, with 99.5784% for the airplane image, 99.4292% for the aerial image, and 99.5784% for the baboon image, demonstrating superior diffusion properties when compared to previous strategies. These high NPCR values show that even minor changes in the plaintext create significant differences in the encrypted pictures, improving security against differential attacks. The proposed scheme has UACI values of 33.5873%, 33.4749%, and 33.3745%, indicating severe confusion with high-intensity differences between encrypted images, hindering efforts to deduce information about the original plaintext. In comparison, enhanced LSB (ELSB) and LSBM approaches offer somewhat lower NPCR values. DCT-based steganography has the lowest NPCR for the airplane at 98.9415%, indicating weaker diffusion [68].

6. Conclusions

The integration of steganography and cryptography within a multi-layered security framework offers a robust method for covert communication, significantly enhancing data protection against unauthorized access. This study presents an advanced crypto-stego model that utilizes AES encryption, DNA sequence encoding, QR codes, and LSB steganography to securely embed text within images. By employing this comprehensive approach, data security is strengthened, ensuring confidentiality, integrity, and resistance against detection. Experimental results demonstrate that the proposed framework effectively safeguards sensitive data while maintaining invisibility and data integrity through sophisticated steganographic methods. The entropy and standard deviation analysis further reveal the impact of various steganographic techniques on information content and texture, with LSB and adaptive LSB methods offering an optimal balance between security and detectability. Additionally, the algorithm showed strong resistance to differential cryptanalysis, achieving high values (99.5784%, 99.4292%, and 99.5784%) and UACI values (33.5873%, 33.5149%, and 33.3745%), confirming its robust diffusion and confusion properties.

While this security model significantly advances secure data transmission, we acknowledge that cryptographic and steganographic techniques can be misused for illicit activities, such as concealing unauthorized communications. To mitigate these risks, implementing advanced forensic analysis techniques, including machine learning-based steganalysis and network traffic monitoring, is essential for anomaly detection. Furthermore, strict access control mechanisms and regulatory frameworks should be enforced to promote ethical and legal compliance. On the other hand, the legitimate applications of our framework are extensive, spanning secure business communications, confidential medical data transmission, intellectual property protection, and secure communications in high-risk environments. By responsibly implementing these technologies, organizations, and individuals can enhance data security while adhering to ethical considerations. Future work will focus on refining the computational efficiency of multi-layered security techniques and developing advanced detection mechanisms to balance security with forensic analysis. By continually adapting these methods, the evolving landscape of cybersecurity can ensure that cryptographic and steganographic advancements serve both security and ethical imperatives in the digital age.