Video Steganography Using a 9-D Chaotic System and an Encryption Algorithm for Data Hiding

The swift advancement of communication technologies and mobile applications has significantly increased the demand for protecting both transmitted and stored information. Consequently, several studies have explored different encryption and steganographic techniques designed to enhance data security.

In 2023, Kunhoth et al. [1] conducted a comprehensive review of video steganography techniques suitable for compressed and uncompressed data. Their analysis included techniques such as least significant bit (LSB) substitution, discrete wavelet transforms (DWTs), and discrete cosine transforms (DCTs). They also explored video steganalysis methods and metrics for evaluating performance. The review provides a comparative assessment of contemporary steganography techniques, emphasising their imperceptibility, robustness, and capacity. It addresses various challenges, such as security vulnerabilities, trade-offs between robustness and capacity, and vulnerability to steganalysis attacks.

In 2020, Guan et al. [2] proposed a novel technique for commutative encryption and data hiding designed explicitly for high efficiency video coding (HEVC) video. The method facilitates the simultaneous encryption and concealment of data while maintaining the original bit rate. The technique emphasizes the selective encryption of quantized transform coefficients (QTCs) and motion vector differences (MVDs) while modifying QTC values to accommodate data embedding. It effectively preserves visual quality, increases embedding capacity, and maintains compatibility with the HEVC standard. This allows for the extraction of the encryption status. However, it depends on strong encryption keys for security, incurs a significant computational overhead, and is limited to HEVC video formats, making it impractical for older video files.

In 2022, Kou et al. [3] introduced a 7-D complex chaotic encryption system that uses a cubic memristor specifically engineered for secure data transmission within innovative grid applications. This approach leverages a high-dimensional chaotic framework to generate robust encryption keys exhibiting high unpredictability. It ensures substantial randomness, is strongly resistant to cryptanalysis, and has an extensive key space, making it exceptionally secure for safeguarding smart grid data. Nonetheless, the methodology demands significant computational resources, restricting its applicability in low-power devices. Moreover, chaotic encryption systems may face synchronization challenges, potentially resulting in data loss.

In 2021, Albahrani et al. [4] performed a comprehensive examination of encryption techniques that use chaotic maps to safeguard audio transmissions. Their findings indicate that these encryption methods offer improved security thanks to their extensive key spaces and sensitivity to initial conditions. A comparative study of alternative techniques confirmed the superiority of chaotic systems in foiling unauthorized access to audio information. Nevertheless, these methods lack stability regarding large audio files, sufficient resilience against hybrid attacks, and struggle to balance security with real-time performance.

In the same year, Vivek and Gadgay [5] introduced a video steganography system based on chaotic encryption, incorporating an enhanced LSB (ELSB) method to securely embed data in HEVC videos. The algorithm applies logistic and Hénon chaotic maps and ensures data integrity before compression. The system achieved a peak signal-to-noise ratio (PSNR) of 35.81 dB with an execution time of 40 seconds, outperforming the common H.264 compression technique (32.72 dB, 45 sec) in imperceptibility and security. However, it has a high computational complexity due to chaotic encryption and requires extended processing times for LSB embedding, making it less efficient for real-time use.

In 2023, Salunke et al. [6] proposed a 3D Gauss encryption compression technique using multiple chaotic keys and Lyapunov exponent validation to enhance security and resistance to brute-force attacks. Despite its enhanced encryption strength and compression, the decryption process demands significant computational resources.

In 2024, Fadhil et al. [7] introduced a method for text encryption and embedding that uses Salas20 encryption alongside Harris corner detection to improve both security and imperceptibility. Although this approach outperforms traditional LSB techniques, its dependence on texture-based embedding could limit its effectiveness in low-texture videos.

In 2024, Meng et al. [8] introduced a secure and efficient coverless video steganography technique that relies on inter-frame similarity. This system creates a secret communication video database (SCVD) using publicly available video data. The SCVD organizes videos according to the similarity of frames between their start and end points using time-based video properties. A mapping table links particular video sequences from the SCVD with segments of confidential information. This mapping serves as a standard reference for both the sender and the receiver, facilitating the embedding and extraction of data without modifying the content of the video. By removing the need for extensive auxiliary data and preventing alterations to the cover medium, the system improves security and maintains the covert nature of coverless steganography. Experimental findings indicate that the proposed method provides enhanced security, capacity, and robustness compared to existing techniques, effectively mitigating the risks of transmitting substantial additional information prevalent in other coverless video steganography methods.

In 2024, Kale et al. [9] presented an innovative video steganography system that secures information by invisibly embedding text within video frames and audio streams. The approach employs unstructured frame selection methods and audio steganography techniques to expand data storage capabilities. Security is further improved using MD5 hashing, which preprocesses the text before embedding. This comprehensive methodology improves both the steganographic process's quality and the embedded data's security, ensuring high imperceptibility. Experimental results show that the method maintains audio and video quality standards while significantly enhancing data embedding capacity and resisting unauthorized access.

In 2024, Dai and Liu [10] tackled a security flaw in the AES algorithm arising from correlations among round keys. Their research employs chaotic systems, including Logistic and Tent maps, to generate highly random sequences that improve AES security. These sequences are used to produce round keys that reduce inter-key correlation. The enhanced key expansion algorithm ensures an ample key space and generates unique encryption keys per session, adhering to the “one secret at a time” principle. Experimental analysis shows that the proposed model eliminates key dependencies, generates robust random sequences, and strengthens resistance to cryptographic attacks.

In their 2024 study, Harsha and Gopika [11] have highlighted the essential role of image protection methods in safeguarding sensitive medical information, military communications, and personal digital data. This review article presents an in-depth examination of modern image encryption techniques. It assesses both traditional symmetric and asymmetric algorithms and more advanced methods such as chaos-based, quantum-based, and hybrid systems. The paper aims to provide researchers and practitioners with a comprehensive understanding of existing encryption methodologies to facilitate the advancement of robust and effective technologies. As a review, it consolidates insights from previous research rather than offering experimental findings, and it critically assesses the advantages and limitations of various encryption strategies.

In 2025, Badhan and Malhi [12] introduced a novel data security architecture that combines various cryptographic techniques to efficiently and comprehensively secure data. Their innovative system uses AES for data encrypting due to its speed and reliability. Elliptic Curve Cryptography (ECC) is used to secure the AES keys, capitalizing on ECC’s lightweight structure and robust secure key exchange mechanisms. To achieve steganographic concealment, the architecture employs an LSB technique, which allows for embedding encrypted data within image files. Furthermore, the system incorporates WebP compression to minimize storage requirements and enhance data transmission efficiency. A multi-tiered security framework is established to meet the demands of contemporary secure communication networks. Experimental evaluations reveal impressive performance, with a PSNR of 68.90 and a Mean Squared Error (MSE) of 0.0083, signifying exceptional image quality following data embedding.

In 2024, Shetty et al. [13] introduced a comprehensive data security framework integrating cryptographic and steganographic methods to achieve efficiency and total confidentiality. The system utilizes AES for the primary data encryption due to its speed and reliability. At the same time, ECC is employed for encrypting AES keys, leveraging its lightweight design and robust key exchange security features. Encryption is further enhanced by applying steganography, specifically the LSB technique, effectively hides encrypted information within image files. The framework integrates Web compression to optimize storage space and improve transmission efficiency. Experimental findings validate the success of this layered strategy, demonstrating a PSNR of 68.90 and an MSE of 0.0083. This indicates negligible distortion and superior image quality post-embedding.

In 2024, Verma and Kumar [14] developed a quantum image encryption system that produces highly unpredictable pseudorandom sequences using a 3-D Bernoulli–Nahem–Map (3D-BNM) chaotic system. This innovative method integrates SHA-256 hash functions to enhance security. The framework converts traditional images into quantum images via the novel enhanced quantum representation (NEQR) format. It subsequently applies spatial scrambling through the generalized quantum Arnold transform (GQAT), followed by qubit-level controlled diffusion facilitated by C-NOT and CC-NOT quantum gates. Experimental evaluates show that this system demonstrates strong resistance to both statistical and differential attacks, all while preserving a low level of computational complexity.

In 2024, Khan et al. [15] introduced an innovative one-dimensional cosine-modulated-polynomial chaotic map to enhance image encryption techniques within Internet of Things (IoT) frameworks. This newly proposed chaotic map demonstrates an expanded chaotic range and possesses key attributes such as aperiodicity, ergodicity, heightened sensitivity to initial conditions, and lower structural complexity. The system successfully met all 17 NIST randomness criteria, and experimental evaluations such as bifurcation diagrams, Lyapunov exponents, and Kolmogorov entropy all validated the map’s robust chaotic properties. These findings underscore its potential utility as a robust pseudorandom number generator for cryptographic applications.

Table 1 provides a systematic comparative analysis of the latest video steganography and encryption developments. Each row details the methodologies utilized alongside their key advantages and limitations. This comparative assessment aids in pinpointing current research gaps and guides the development of future studies. Compared to existing work, our scheme involving a Nine-Dimensional Chaotic Logistic System with AES encryption presents a new algorithmic technique for generating highly random keys that can enormously reduce the cryptanalysis. An optimized LSB embedding on video frames leads to perfect RD, while a rare combination of large capacity and low distortion is achieved.

Table 1. Comparative analysis of recent video steganography and encryption techniques

This section outlines the framework of the proposed system, which incorporates various algorithms, such as steganography, AES, and a 9-D chaotic system. These methodologies work in tandem to bolster security and ensure resilience against potential attacks by leveraging the advantages of established cryptographic techniques. Figure 1 illustrates the framework.

1.png

Figure 1. A hybrid framework for the proposed system

4.1 Key generation mechanism based on the 9-D chaotic logistic system

The 9-D coupled chaotic logistic system exhibits complex chaotic behavior, making it a strong candidate for cryptographic key generation. The system is defined using nine coupled logistic maps.

4.1.1 The 9-D logistic system

The following equation governs the behavior of this system:

$x_i^{t+1}=r_i \cdot x_i^t \cdot\left(1-x_i^t\right)+\epsilon \cdot \sum_{\substack{j=1 \\ i \neq i}}^9(-1)^{i+j} x_j^t$           (1)

where,

$x_i^t=$ the state of the $i$, the logistic map at time step $t ;$

$r_i=3.99$ the chaos parameter, whose value ensures the system operates in a fully chaotic regime;

$\epsilon=0.015$ the coupling strength, which links all maps together.

The initial values are taken to be

$\begin{gathered}X=[0.123456789,0.234567890,0.345678901, \\ 0.456789012,0.567890123,0.678901234, \\ 0.789012345,0.890123456,0.987654321] .\end{gathered}$

These values are chosen to avoid falling into fixed points or periodic orbits. A burn-in phase of 1000 iterations is used to eliminate transient effects. For key generation, 111 112 iterations are performed after the burn-in phase. The outputs are binarized with a threshold of 0.516. The threshold of 0.015 was chosen after testing various values. It provided the best balance between detected units and false alarms. Lower values caused excessive alerts, while higher ones missed important data. Thus, 0.015 ensured better accuracy and stability.

4.1.2 Chaotic key generation process

Keys are generated according to the following procedure.

1. System initialization:

The nine logistic maps start from predefined initial conditions.

Burn-in iterations are performed to remove transient effects.

2. Iterative chaotic map computation:

At each iteration, all $x_i$ values are updated using the logistic equation and coupled interactions.

3. Binary key extraction:

The output values are compared with a threshold (0.516).

If $x_i \geq 0.516$, then the corresponding bit is set to 1; otherwise, it is set to 0.

The system achieves high entropy and generates a truly random-like sequence. Since the system operates in nine dimensions, the generated keys are taken from an ample space, making brute-force attacks impractical. In addition, it is irreversible without the exact knowledge of the initial conditions.

4.2 Visual analysis of the chaotic system

Figure 2 shows a time-series representation of the 9-D logistic map over multiple iterations, demonstrating its chaotic behavior. Each panel corresponds to a dimension. The wide distribution of values from 0 to 1 confirms the unpredictable nature of the system.

2.png

Figure 2. Chaotic behaviors of nine logistic maps

4.2.1 3-D representation of chaotic behavior

To illustrate the interaction between the variables in the chaotic map, Figure 3 shows a 3-D phase plot for the first three. The scattered distribution of points confirms the nonlinear and nonperiodic nature of the chaotic system and demonstrates its suitability for cryptographic key generation.

3.png

Figure 3. 3-D phase plots of the chaotic system

4.2.2 Distribution of chaotic values

The histogram in Figure 4 illustrates the distribution of values generated by the 9-D logistic map. The relatively uniform spread, with peaks at the extreme values (0 and 1), confirms the system's high entropy, ensuring strong randomness for key generation.

4.png

Figure 4. Histogram distribution of logistic maps

4.3 Applications in cryptography

The extracted binary sequence is formatted into 256-bit AES keys for AES key generation. The keys are used for secure data hiding in video frames. This ensures high-security key generation with enhanced unpredictability and robustness.

4.3.1 Secure data hiding and steganography in video using the AES and LSB algorithms

The text encoding and decoding process involves converting the ciphertext from base 64 to binary data using AES encryption. The information is divided into two segments: the initialization vector and the encrypted information. The initialization vector is used to create an AES cipher object, which is decoded using AES decryption. The video is subsequently segmented into frames and saved in a designated output directory. The steganography procedure entails a statistical examination of the byte lengths of headers, which indicates the duration of concealed records. The LSB method is employed to conceal the information, with the hidden records dispersed across various frames and segmented into bytes for secure integration. The length of the initial frame is established by summing the first four bytes of concealed data. The hidden text is retrieved from the video, analyzed frame by frame, and reassembled into an image file.