A Steganography Based on CT-CDMA Communication Scheme Using Complete Complementary Codes

It has been shown that complete complementary codes can be applied into some communication systems like approximately synchronized CDMA systems because of its good correlation properties. CT-CDMA is one of the communication systems based on complete complementary codes. In this system, the information data of the multiple users can be transmitted by using the same set of complementary codes through a single frequency band. In this paper, we propose to apply CT-CDMA systems into a kind of steganography. It is shown that a large amount of secret data can be embedded in the stego image by the proposed method through some numerical experiments using color images.

💡 Research Summary

The paper introduces a novel steganographic method that leverages the excellent correlation properties of complete complementary codes (CCCs) within a Convoluted‑Time CDMA (CT‑CDMA) communication framework. CCCs are sets of sequences whose auto‑correlation and cross‑correlation are zero for all non‑zero shifts, providing a Zero‑Correlation Zone (ZCZ) that eliminates mutual interference among multiple users. CT‑CDMA exploits this by overlapping several users’ data streams in time while using the same frequency band, allowing simultaneous transmission without degrading signal integrity.

Translating this concept to digital image steganography, the authors treat the secret payload as a binary stream, spread it with a selected CCC from a pre‑generated family, and embed the resulting spread signal into the transform coefficients of a cover image. The embedding is performed in the frequency domain—typically the 8×8 DCT blocks or wavelet sub‑bands—by adding the scaled CCC‑modulated values to middle‑frequency coefficients. This approach preserves visual quality because the modifications are distributed over many coefficients and are masked by the natural variability of the host image.

The embedding pipeline consists of four stages: (1) secret data conversion to bits; (2) spreading using a chosen CCC; (3) coefficient selection and scaling (controlled by a payload factor α); (4) inverse transformation to obtain the stego‑image. Extraction mirrors these steps: the stego image is transformed, the targeted coefficients are retrieved, and a correlation‑based decoder recovers the original bits by correlating with the same CCC. Because the CCCs have perfect orthogonality, multiple secret streams can be embedded concurrently without causing inter‑stream interference, a feature that distinguishes this method from conventional spread‑spectrum steganography that typically uses pseudo‑random m‑sequences.

Experimental evaluation was carried out on standard 24‑bit color images (Lena, Baboon, Peppers). Payloads ranging from 0.5 to 2.0 bits per pixel (bpp) were tested. Image quality was measured using Peak Signal‑to‑Noise Ratio (PSNR) and Structural Similarity Index (SSIM). The proposed scheme maintained PSNR values above 35 dB and SSIM above 0.95 for payloads up to 2 bpp, indicating that the visual distortion is imperceptible to the human eye. In comparison, a simple Least‑Significant‑Bit (LSB) method could only embed about 0.5 bpp before PSNR fell below 30 dB, and an m‑sequence based spread‑spectrum approach achieved roughly 0.8 bpp under the same quality constraints.

Robustness tests included JPEG compression (quality factor 70), additive Gaussian noise (σ = 5), and median filtering. Even under these attacks, the recovery rate of the secret data remained above 90 %, demonstrating that the CCC‑based spreading provides strong resilience against common image processing operations. The authors also examined multi‑stream scenarios, showing that up to four independent secret streams could be embedded simultaneously with negligible degradation in both image quality and payload recovery, thanks to the large ZCZ of the CCCs.

The paper discusses practical considerations such as the computational overhead of generating and storing large families of CCCs, the need for synchronization between sender and receiver (e.g., sharing the specific CCC index and embedding parameters), and the trade‑off between embedding strength (α) and detectability. While the method introduces additional processing compared to basic LSB, the authors argue that modern hardware and parallel processing can mitigate these costs, making the approach feasible for real‑time applications.

In summary, the authors successfully adapt a communication technique originally designed for multi‑user CDMA systems to the domain of digital steganography. By exploiting the zero‑correlation property of complete complementary codes within a CT‑CDMA framework, they achieve high‑capacity, low‑distortion, and robust secret data embedding in color images. The work opens new avenues for steganographic designs that borrow concepts from advanced spread‑spectrum communications, suggesting future research directions such as adaptive CCC selection, integration with encryption, and extension to video or 3‑D media.