Abstract
Although the development of watermarking techniques has enabled designers to tackle normal processing attacks (e.g., amplitude scaling, noise addition, re-compression), robustness against malicious attacks remains a challenge. The discordant size content replacement attack is an attack against watermarking schemes which performs content replacement that increases or reduces the number of samples in the signal. This attack modifies the content and length of the signal, as well as desynchronizes the position of the watermark and its removal. In this paper, a source-channel coding approach for protecting an audio signal against this attack was applied. Before applying the source-channel encoding, a decimation technique was performed to reduce by one-half the number of samples in the original signal. This technique allowed compressing at a bit rate of 64 kbps and obtaining a watermarked audio signal with an excellent quality scale. In the watermark restoration, an interpolation was applied after the source-channel decoding to recover the content and the length. The procedure of decimation–interpolation was taken because it is a linear and time-invariant operation and is useful in digital audio. A synchronization strategy was designed to detect the positions where the number of samples in the signal was increased or reduced. The restoration ability of the proposed scheme was tested with a mathematical model of the discordant size content replacement attack. The attack model confirmed that it is necessary to design a synchronizing strategy to correctly extract the watermark and to recover the tampered signal. Experimental results show that the scheme has better restoration ability than state-of-the-art schemes. The scheme was able to restore a tampered area of around 20% with very good quality, and up to 58.3% with acceptable quality. The robustness against the discordant size content replacement attack was achieved with a transparency threshold above −2.
Highlights
The development of watermarking techniques has enabled designers to tackle normal processing attacks, robustness against malicious attacks remains a challenge
A total of 150 audio signals were subject to the protection against possible tampering by the discordant size content replacement attack
The % recovery denotes the tampered area percentage that the scheme is capable of recovering. These schemes have been chosen for comparison based on robustness against the discordant size content replacement attack and its recovery percentage
Summary
The self-recovery algorithm proposed was developed on the basis of a small version and presented earlier in [3] The goal of both algorithms is to recover signals that have been tampered with by a discordant size content replacement attack. The channel coding and hash information computed for each frame were used to construct the watermark, which was inserted in the least significant bits (LSB). The hash information helps determine the tampered frames, the channel coding is processed with a channel decoder and a source decoder. This part of the reconstructed watermark is applied to recover the tampered blocks of the host audio signal. The mathematical model of the discordant size content replacement attack was introduced and used to evaluate the restoration capability of the scheme
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call