Content-Aware Quantization Index Modulation for Physical Layer Steganography

Huiyi Tang, Junlong Mao, Tengjian Liu, Shanxiang Lyu

Abstract: Conventional Quantization Index Modulation (QIM) has been tailored for diverse data hiding applications, yet it lacks the ability to achieve minimal distortion in physical layer steganography. To address this, we introduce an enhanced QIM scheme, Content-Aware QIM (CA-QIM). By adapting the codebook to the probability distribution of embedded messages and host signals, CA-QIM excels at handling non-uniformly distributed messages, resulting in significantly reduced distortion. Our approach outperforms other QIM variants in terms of mean square error (MSE), peak signal-to-noise ratio (PSNR), and percentage residual difference (PRD), as demonstrated through simulation results. CA-QIM offers a promising solution for achieving improved perceptual transparency and robustness in steganography applications.

External IDs:dblp:conf/globecom/TangMLL23