Amirkabir University of TechnologyAUT Journal of Mathematics and Computing2783-24493120220201A new approach to solve the reliability problem in any VoIP steganography system113127476910.22060/ajmc.2022.20710.1073ENM.-RSadeghiDepartment of Mathematics and Computer Science, Amirkabir University of Technology, Tehran, Iran0000-0002-7676-4168P.Amirzade DanaDepartment of Mathematics and Computer Science, Amirkabir University of Technology, Tehran, IranB.JavadiDepartment of Mathematics and Computer Science, Amirkabir University of Technology, Tehran, IranJournal Article20211026VoIP is naturally an unreliable communication system. Thus, using the best VoIP steganographic systems, the accuracy of the hidden message is impaired as a result of the VoIP packet loss. There are many steganography and steganalysis researches that try to improve the robustness and accuracy of VoIP steganography methods. In addition to the fact that these works are done depend on a particular method, none of them have solved the problem of packet loss. Applying error correcting codes, prior to embedding, is a well-known technique in telecommunication to improve robustness and to reconstruct Missing data. However, in the case of VoIP communication, a codeword entirely embedded in the packet may be lost due to the packet loss and therefore ECC techniques will not be capable of reconstructing the lost bits. In this paper, we design a novel scheme to increase the reliability of VoIP steganography systems. We emphasize that our proposed method, independent of the embedding and extracting algorithm, can be used in all VoIP steganography systems. After encoding the secret message to the codewords of $n$ bits, we distribute these $n$ bits into $n$ successive RTP packets, in such a way that, losing one packet leads to miss only one bit of each codeword. Then, with the idea of telecommunication solutions in recovering lost data, when up to $t$ of $n$ packets are lost we can recover the secret message using a $t$-error correcting code ${\bf C}(n,k,d)$. Provided that the average of packet loss over the network is less than $1\%$, using a $t$-error correcting code ${\bf C}(n,k,d)$, the probability of losing hidden data, in each category of $n$-packets, $P_e$, is less than $\leq 10^{-2t}$. Hence, applying the $t$-error correcting codes with larger $t$, in the proposed scheme, results in more reliable steganographic systems.https://ajmc.aut.ac.ir/article_4769_696b241fdf88dc6eeb233c7364662e2d.pdf