Covert Communication Schemes Based On The Multiple Description Image Coding

The multiple description coding (MDC) is a kind of method which is designed to resist the time delay in the unreliable channels. Generally the object to be encoded in the MDC system is image, audio, and video etc. So if the covert communication scheme is based on the MDC system, then the covert communication scheme can also have the ability to resist the channel errors and the time delay, and then the purpose of designing a reliable covert communication can be achieved.There have been a lot of papers concerning about the MDC and the information hiding algorithms, while few papers combine the MDC and the information hiding algorithms together, the dissertation compensates this vacancy. In the dissertation, the attention is mainly paid on designing robust information hiding schemes and let the information hiding algorithms be compatible with the MDC systems, so that the covert communication schemes can also resist the noise, packet loss, and fadings in the channels. The main innovative contributions of this dissertation are as follows:Firstly, the hamming distance based index partition algorithm is proposed to improve the performance of the multiple description scalar quantization (MDSQ) based covert communication schemes. Firstly, the carrier image is encoded by the vector quantization (VQ) method, the secret information is embedded in the indices by the code book partition based method during the VQ encoding phase; secondly, the original index set is divided into two equal-sized sub-index sets, and one sub-index set corresponds to one sub-codebook; thirdly, the watermarked indices are sent to the index assignment matrix of MDSQ, one index is mapped to two or more sub-indices, the sub-indices are transmitted through the channels individually. The experimental results show that the index partition algorithm improved the robustness of the secret information during the transmission.Secondly, the covert communication schemes are proposed based on the subsampling-DCT and DWT multiple description coding systems. Firstly, the carrier image is subsampled into subimages. Secondly, the secret information is embedded in the quantized DCT or DWT coefficients of the subimages. Thirdly, the subimages are transmitted through the different channels after JPEG or SPIHT coding. The traditional subsampling based information hiding algorithm can not resist the permutation attack. In this paper, an improved algorithm is proposed to resist the permutation attack. The experimental results show the effectiveness of the proposed algorithm.Thirdly, the DWT and data partition MDC based covert communication scheme is proposed. Firstly, the carrier image is decomposed by the DWT, and then the horizontal, vertical, and the diagonal DWT coefficient sets are SPIHT encoded respectively. The secret information is embedded in the low frequency DWT coefficients by the parity modulation based method.Fourthly, the transform and spatical domains based information hiding algorithm is proposed, the performance of the double domains based information hiding algorithm is tested under the FEC based MDC system. At the transmitter side, the secret information is embedded in the DWT domain and the pixel domain respectively. At the receiver side, the secret information is extracted twice, one is in the pixel domain, the other is in the DWT domain, select the secret information image which has the better visual quality as the final result. The experimental results show that the information hiding algorithm can resist not only the common image processing attacks, but also the geometrical attacks.Lastly, a phase modulation based information hiding is proposed for the Orthogonal Frequency Division Multiplexing (OFDM) system in the signal mapping phase. The proposed information hiding algorithm has no relationship with the carrier, the experimentals results show the effectiveness of it.