Selective Multimedia Data Encryption Based on Discrete Hyperchaotic S-Box Generation with Motion-Based Fragmentation

doi:10.21203/rs.3.rs-1418622/v1

Download PDF

Research Article

Selective Multimedia Data Encryption Based on Discrete Hyperchaotic S-Box Generation with Motion-Based Fragmentation

https://doi.org/10.21203/rs.3.rs-1418622/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Cloud computing has become the significant technology with rapid popularization offering many services with highly effective computation and large scale storage solutions. As the multimedia files are growing explosively, securely storing it for authorized access is the major task. Concentrating on video file storage in public cloud environment, security and privacy of those outsourced videos are major concerns. The semantic elements of the HEVC videos are extracted on the horizontal and vertical video fragments and Selective Encryption on those semantic elements of videos are done when the data owner uploads video for cloud storage. Modified AES is the proposed selective encryption scheme to protect the video by encrypting each fragments in cipher feedback (CFB) process mode. A dynamic, nonlinear 2D discrete hyperchaotic mapping based S-box generation is proposed to enhance the security level by providing good confusion, diffusion and scrambling. The performance analysis of generated S-box and experimental results shows that the proposed selective encryption highly resists attacks and encrypts with reduced time complexity.

selective encryption

semantic elements

discrete hyperchaotic mapping

nonlinear S-box

Various approaches of cloud technology, cloud services have reached a tremendous progression of computational clouds with constrained resources. As cloud environment has grown tremendously with numerous services such as infrastructure as a service (IaaS), platform as a service (PaaS), software as a service (SaaS), security as a service, conventional encryption techniques discussed in [1] need to be enhanced to deal with current emerging security threats. Cloud users outsource their sensitive confidential information to cloud server. Major tasks of deploying computational cloud services are to enhance security, reduced utilization of resources, minimize delay. Though many traditional encryption standards are developed but still there exists security issues and privacy concerns. Many existing techniques consume more time to encrypt and decrypt data blocks and also less security exists in stream ciphers.

Encrypting a video is a time consuming process as video file consists of huge amount of information. Video encryption can either be full (naïve) encryption or selective encryption. Those video file comprising smaller size can be encrypted completely which is known as full (naïve) encryption. The larger sized videos can be selectively encrypted on extracting the semantic elements and encrypting it. The more informative semantic elements of the video are Intra prediction mode, inter prediction mode, non-zero coefficients of the Intra coded (I) frame in addition predicted (P) frame, Motion Vector (MV), motion vector difference (MVD) of macroblocks, quantization para6meters difference (QP) and the residual coefficients. Primary goal of video encryption is that it must satisfy the necessary conditions such as less computation complexity, less cost in time, no format compliance and should not influence compression efficiency. Shah et al., (2020)[4] encrypted syntax elements of the HEVC and SHVC videos selectively. Though this method protects the visual contents, it is not suitable for real time applications and there is need of security enhancement. He et al., (2020) [8] selectively encrypted the non zero coefficients of DCT sign bits but it does not resist differential attack.

A lightweight encryption of surveillance videos is deployed by (Zhang et al., 2018) [10] where the region of interest content is protected based on cellular automata layered approach. From the group of pictures (GOP) of video, ROI of I frame is alone encrypted and so there is leakage of data and is vulnerable to attack. Ma et al., (2015) [23] has also protected the ROI of advanced video coding (AVC) / H.264 compressed videos. Due to quantization process the privacy content are distorted during encoding. Shah et al., (2019) [11] has performed transcoding and decoding of video using XOR operation to encrypt and has not concentrated on syntax elements and rate distortion quality is not analysed. DNA based encryption scheme is proposed by (Namasudra et al., 2020) [19] to encrypt the multimedia files uploaded to the cloud environment but the computation cost and latency are high.

A selective encryption technique is proposed in this paper, where the pseudo random number generator (PRNG) is used to generate key; then AES in cipher feedback (CFB) mode is applied on the extracted MVD, IPM, QP elements to encrypt it. A modified 2D discrete hyperchaotic algorithm is proposed to generate a S-box (substitution box) dynamically in AES-CFB operation to strengthen the security of video encryption by increasing nonlinearity and decreasing differential probability. Security of the Advanced Encryption Standard (AES) is enhanced by modifying the AES by using chaotic maps for substitution box generation and the speed of encryption process is increased.

The organization of the paper is shown as follows: section 2 details the preliminaries of HEVC video encoders as well as video encryption. Section 3 defines the functionality of chaotic maps. The proposed selective encryption with hyperchaotic based S-box creation is elaborated in section 4. In section 5, experimental result analysis with its performance metrics is discussed. Finally in section 6 conclusion is given.

HEVC encoders use quadtree structure of block partitioning with large inter/intra prediction blocks, efficient coding architecture and high bit depth. HEVC encoder uses context adaptive binary arithmetic coding (CABAC) encoding. It involves mainly three functions such as binarization step, context modelling and arithmetic coding. In binarization, all the non-binary semantic elements of the video is transformed into binary symbols or strings (bin). The probabilities of bins are updated during the context modelling phase. Based on these probabilities, the bins are compressed to the bits. During arithmetic coding, based on the estimated probability, bins are compressed to bits in two different modes such as regular mode or bypass mode. In regular mode, the probability is selected from the content based on previously encoded syntax elements whereas in bypass mode, fixed probability is used. Ensuring format compliance and maintaining the bit rate HEVC selective encryption is performed by using the bypass mode [2].

Non zero coefficients and sign bits of Quantized Transform Coefficients(QTC) are encrypted using stream cipher by performing XOR operation among plaintext and PRNG secret bits[2]. The most important part in video construction is motion vector (MV). Instead of encrypting MV, the signs and magnitude of motion vector difference (MVD) are coded in bypass mode and encrypted. The suffix value of the MVDs are XORed with the Pseudo random bits. Performing encryption of bins in bypass mode (such as sign bits of QTC, MVD and absolute suffix value of MVD) ensures format compatibility of video and thus maintains structure of video in achieving the same bit rate. In HEVC, the first two bins are encoded in regular mode and the remaining are encoded in bypass mode by employing Golomb-Rice and Exponential-Golomb (Exp-Golomb) for small values and large values respectively. Hong et al., (2014) [22] extracted the MV and encrypted with RC4 stream cipher approach which has format compliance and increase in bits.

2.1 Fragmentation of video

In early days, video content analysis and extraction are performed researchers mainly focussed on video cut detection techniques. Later the recent works focussed on detecting the transitions and other techniques. To detect the shot in the inputted video, the following techniques were employed.

The most common and simplest algorithm is determining the sum of absolute pixel difference. Here, Pixel wise comparison is performed among two successive frames.
In histogram comparison, the disparity amongst the histograms of two successive frames are computed.
Region of interest (ROI) is considered from the image and then further pixel difference is computed statistically in ROI.
Edge difference, which determines the number and position of the edge. The original information of two frames is compared by means of converting both frames into edge images.
Threshold comparison is done by determine the threshold value of current frame and comparing it to the next frame to detect the shot.
Motion vectors are detected using block matching algorithms. Horizontal and vertical fragmentation can be done on the video files based on size and time.

2.2 Traditional Encryption Algorithm

Encryption technique provides the confidentiality of the outsourced data to the public clouds and protect the data from adversary by applying distinct numerical systems and symmetric/asymmetric key to transform digital information into meaningless ciphertext. The objective of security management is to provide data confidentiality, authentication of users / devices, integrity and accuracy. Moreover, video-based as well as image-based files requires more time and power consumption for encryption as well as decryption. Although it is recognized that statistics multimedia security gives extra blessings, protection and confinement, but the concerns raised due to vulnerable access of customer records. There are many security issues related to customers data. Traditional encryption offers confidentiality, but it is not well suited for multimedia security. In recent encryption methods, users exclusively encrypt their data with their own keys providing increased security. This involves the following steps:

Key_Gen; is the key generation procedure, which creates key k using PRNG or chaos parameters.

Encrypt (k, pt): is the encryption procedure that encrypts or translates the plain text, pt with the help of key k and then generates the cipher text, ct in unreadable form.

Decrypt (k, ct): is the decryption procedure that decrypts the cipher text, ct into plain text, pt of original form with the help of same key k.

Tag_Gen(M): is the tag generation algorithm that outputs a tag value T for the message M which is useful to check the integrity of M.

Advanced Encryption Standard (AES) algorithm is based on the principles of substation and permutation. AES reads 128 bit sized plaintext chunk and encrypts it using various sizes of keys 128 or 192 or 256 bits for 10 or 12 o4 14 rounds of operation respectively. This plaintext is replicated to the state array that is updated in each round and at the last phase, it is copied to the output. Each round performs mainly four different operations such as substitution bytes(S-box), shift rows, mix columns and add round keys. These are the 4 stages of round transformations involving one permutation and three substitution functions. Initially encryption begins with one add round key operation using the key. Add round key performs bitwise XOR of current block with 4 words of expanded key. Substitute byte stage uses substitution box (S-box) to perform substitution of bits in a block. Followed by a simple permutation in shift rows stage. Output of this stage is then passed to mix column, substitution is performed using arithmetic multiplication over GF(2⁸). These 4 stages of operations are performed for n-1 rounds and then mix column is eliminated in nth round. All the stages provide confusion, diffusion, nonlinearity and so it is secure. All these stages are reversible and so the decryption uses expanded ke in reverse order.

AES encryption algorithm can be implemented in four different modes of operations such as, Electronic codebook (ECB), cipher block chaining (CBC), cipher feedback (CFB) and counter (CTR) mode. In this work, CFB method is preferred to encrypt the video fragments in a bit sliced approach. The same bit rate is accurately maintained by CFB and also produces format compatibility. CFB mode divides the plaintext, pt into k bit segments rather than m (128) bits. An initialization vector (IV) is initialized which is m-bit shift register. It is similar to stream cipher approach. The leftmost most significant bit (MSB) is k bits of encryption function output which is being XORed together with the initial part of plaintext pt1 towards generating ciphertext ct1 and is transmitted. Meanwhile the content of shift register is moved towards left for k number of bit positions and ct1 is restored in rightmost least significant bit (LSB) of k bits in the shift register. This procedure is repeated till the entire plaintext blocks are encrypted. The input register is renewed as one block at a time with feedback structure. CFB produces output depending on both plaintext and ciphertext.

A chaotic system, a mathematical model consisting of more than one positive Lyapunov Exponent (LE) is said to be hyperchaotic. Hyperchaos is the new class of attractors raised with 4 dimensions as a minimum. This hyperchaotic attractor has atleast two positive LE showing chaos behaviour and one negative LE ensuring the boundness of the solution. These hyperchaotic systems have complex algebraic structure which further makes difficult to obtain the bounds. The dynamic complex behaviour of chaotic systems can be applied in the field of dynamics, cryptosystems, circuits, engineering, biochemical responses, stock exchanges, etc. researchers have concentrated more on using hyperchaos behaviour to increase nonlinearity. A new state variable can be introduced to the 3-D chaotic system to make as hyperchaotic in a simple way.

Cheng et al., (2020) [9] employed 4D hyperchaotic system, where the extracted syntax elements from each slice of video is first encrypted with the keys generated by PRNG. These encrypted elements are then inputted to the 4D hyperchaotic system performing DNA code selection and random matrix coding in the codeword substitution phase. Seven reference sequences from the group of pictures consisting of 300 frames are encrypted perceptually with four different sequences generated by chaos system. Chaos based multimedia encryption proposed by (Yasser et al., 2020)[7] with perturbation technique is included in all chaotic maps based on Linear Feedback Shift Register (LFSR) consisting of 16 units where each of the unit has 31 bits. Sallam et al., (2018) [13] encrypted the MVD sign bits and DCT coefficients with chaotic logistic map. 2D logistic map use high computational cost. Qiao et al., [5] paired the Chebyshev discrete chaos map and skew tent map to create the random sequence for introducing more confusion in S-box. Hamming distance and correlation analysis are measured to evaluate the confusion and diffusion property in image encryption. García et al., (2018) [14] suggested a simple D chaos based image encryption but the generated S-box is weak due to less nonlinearity and hence does not resist linear attack. Logistic sine chaotic maps and combined with DNA sequence based permutation to encrypt the image by (Chai et al., 2019) [21]. Though the proposed scheme generates DNA matrix and perform XOR operation to resist the known attacks, but it does not provide high nonlinearity with good diffusion.

Image encryption scheme is the foundation of video encryption techniques performed frame by frame manner. Some authors have used the chaotic map directly on the image to encrypt it. Based on chaotic map, large pseudo random bit sequences are generated and then XOR operation is performed on these bit sequence and the plain text stream to encrypt it. Valli et al.,(2017) [16] implemented a pipeline for encrypting videos using S-Box with high dimension of chaos and delayed differential equation. But there exists key complexity and more time complexity for encryption. Zhu et al., [6] proposed a S-Box generation using combined chaos algorithms. It is then optimized based on the genetic algorithm involving mutation and crossover stages. Hussain et al., (2019)[12] constructed the substitution box based on mobius transformation, logistic map to provide nonlinearity resisting to known attacks. Malik et al.,(2020)[3] has developed a dynamic S-box using a simples logistic chaotic map and the resulting S-box is then combined with affine transformation to perform rotational operation during key scheduling so as to increase the nonlinearity. The proposed work enhanced the AES security but with limited S-boxes. Hamidouche et al., (2017) [15] and (Taha et al., 2017) [17] has deployed skew tent combined with piece wise linear chaos map to generate pseudo random sequence generator and encrypt the syntax elements of the video in block cipher and stream cipher approach respectively but the computational cost was high.

A secure cloud framework is designed where the videos uploaded by the data owner or cloud user is securely stored in a block cipher feedback mode. The outsourced video data is first fragmented into several fragments and then each fragment is encrypted using the key pair generated at owner side using modified AES. The existing approach of fragment-then-encrypt provides security but there is an increased risk of data loss. The proposed framework of enhancing secured storage of videos in cloud is shown Fig. 1. The video data uploaded by the owner is fragmented horizontally / vertically and then encrypted using modified AES with 2D discrete hyperchaotic map based S-box. In CFB mode of AES operation, the current plaintext block is being XORed with the ciphertext of previous block to create a current ciphertext block. During the initial phase, plaintext is XORed with the initialization vector to generate the first ciphertext block of 128 bits.

Encryption of video files are performed as

Step 1) Data owner upload the video files for secured storage with protected access. This outsourced video file is fragmented into many fragments in frames in temporal and spatial frame space. The video file is passed to Fragmentation module with the parameters such as target_file_path, number_of_fragments, output_location performs motion based fragmentation which then produces sequential fragments with sequence ID.

Step 2) Semantic elements of the input video such as codewords of Inter/Intra Prediction Mode (IPM), Moving vector Difference (MVD), Signs of residual coefficients are extracted from these fragments using context adaptive binary arithmetic coding.

Step 3) 2-Dimensional hyperchaotic operation is performed on the binary strings involving permutation and diffusion functions.

Step 4) Chaotic mapping is performed to generate a adaptive encryption key generation and thus generates a chaotic sequence of S-Box.

Step 5) Cipher Feedback (CFB) mode of AES with the adaptive dynamic key is used to encrypt each slice and Galois Counter Mode (GCM) for authentication code is used to preserve the integrity.

Step 6) Objective video quality analysis is performed by measuring Peak Signal to Noise Ratio (PSNR), Structural Similarity Index (SSIM) and Information Entropy value.

AES algorithm in cipher feedback (CFB) mode of operation is performed to selectively encrypt the extracted syntax elements. As the S-box in AES is vulnerable to timing attack, a modified dynamic S-box is generated using 2D discrete hyperchaotic mapping. Chaos signals are dynamic which pose a good characteristic of pseudorandom sequence. This chaotic system is highly complex, unpredictable which produces good randomness and highly nonlinear. Figure 2 shows the modified AES encryption algorithm.

Substitute byte operation in AES involves chaos based S-box which process hyperchaotic behaviour. The proposed modified AES with 2D-dicrete hyperchaotic system has been simulated and tested with different parameters.

4.1 S-Box generation

The algorithm to create a dynamic nonlinear S-box involved in AES is outlined as follows:

Step 1: Initialize the parameters of the 2D discrete hyperchaotic map.

Step 2: Iterate the map and generate chaotic sequences ${x}_{i}$ and ${y}_{i}$ based on the defined mapping equation.

$${x}_{i}=r\left(\text{sin}\left(\pi {y}_{i-1}\right)+a\right){\bullet x}_{i-1}\left(1-{x}_{i-1}\right)$$

$${y}_{i}=r\left(\text{sin}\left(\pi {x}_{i}\right)+a\right)\bullet {y}_{i-1}(1-{y}_{i-1})$$

Where r is the control parameter which shows the hyperchaotic actions while its value remains in range of [0.904,1], a = 3, ${x}_{0}=0.11$ and ${y}_{0}=0.14$.

Step 3: Generate a random sequence between 0 to 255 (matching length of S-box) by converting xi.

$${P}_{i}={floor(x}_{i}\times {10}^{6}) mod 256$$

Step 4: Sort the sequence ${y}_{i}$ and make a note of its location as an index sequence Q.

Step 5: Select the value in P that corresponds to Q and make sure if it already appears in the S-box. If not, insert the P value to the S-box until its length (256) gets filled with unrepeated values. 1024 S-boxes are randomly generated.

Step 6: Scramble the generated S-box.

Hence S-box is generated based on above mentioned steps which have high efficiency and nonlinear. The strength of AES cryptographic algorithm depends on the strong S-box that resists against known attacks. The S-box is reversible and hence during the decryption process, inverse of S-box is applied with inverse diffusion and inverse scrambling. The S-Box strengths are analyzed which depends on algebraic structure nonlinearity, bijective, strict avalanche criterion (SAC), Bit Independence Criterion (BIC), linear probability and linear and differential cryptanalysis. There are 1024 boxes generated with 112 nonlinearity. Therefore, more randomness are produced from the nonlinear S box.

The integrity of the block transmitted are verified using Galois Counter mode (GCM) which is parallelizable providing high throughput with less latency and low cost. It allows arbitrary length of plaintext bits strings. It includes two functions namely, Ghash and CTR. GCM is mentioned for the purpose of authenticated encryption in a stream ciphering approach. As the proposed framework uses CFB block cipher mode of modified chaos based AES, the GCM uses only GHASH function to include authentication tag for the transmitted data. It outputs a 128 bit sized tag. At the receiver end, this received tag can be used to check whether the data received is tampered or not by comparing it with a new computed. GHASH takes input plaintext 128 bit string, $pt$ and hash key and then performs multiplication in Galois Field.

Let ${pt}_{1}, {pt}_{2}, {pt}_{3}\dots {pt}_{m}$ denote the unique block sequences such that

$$pt={pt}_{1}\parallel {pt}_{2}\parallel {pt}_{3}\parallel \dots \parallel {pt}_{m}$$

Let ${t}_{0}$ be block of 128 zeros. For all values of i = 1 to m, compute the tag as

$${t}_{i}=({t}_{i-1}\oplus {pt}_{i})\cdot H$$

Where H is the hash key. This then returns the ${t}_{m}$ tag value which is appended to the cipher block.

The proposed selective encryption of videos is implemented in a cloud application by creating microservices adding frontend web components using ReactJS, a cloud native technique based on Go language. The simulator was developed using Cloud Native React as Front End with the support of library System Security which is implemented within JavaScript and using the default settings. The performance analysis is done on the created S-Box by measuring, nonlinearity, SAC, BIC, approximation of differential and linear probability. Differential properties of S-box are analysed as it is intended to resist differential attacks. To evaluate the objective video quality, Mean Square Error (MSE), Information Entropy value, Peak Signal-to-Noise Ratio (PSNR), and Structural Similarity Index (SSIM) are measured for the encrypted frame.

5.1 Nonlinearity

Nonlinearity of a function f(x) can be described as the least possible hamming distance among the functions f and H, where H represents the collection of the entire linear functions in addition to affinal functions. Here f(x) is a Boolean function of Galois field GF(2ⁿ).

$${N}_{f\left(x\right)}={min}_{h\in {H}^{n }}\text{d}(f,h)$$

Higher the value of nonlinearity indicates stronger the S-box which resists linear attacks. The proposed chaotic s-box produces nonlinearity as 112 which is efficient of resisting nonlinear attacks.

5.2 Strict Avalanche Criterion

The avalanche effect of the Boolean function is quantitatively analysed by the SAC [18]. It can be defined as one bit difference in the given input Boolean function must alter atleast one-half of the of the output bit values (i.e. changing probability should be 0.5). Dependency matrix or correlation is calculated to evaluate the SAC in which each element of the matrix will be 0.5. The average of all the elements in matrix will be closer to 0, indicating that it satisfies SAC.

5.3 Bit Independence Criterion (BIC)

Let the two distinct Boolean function be ${f}_{i}{ and f}_{k}where \left(i\ne k\right)$ and these two are the outputs of the S-box with high nonlinearity. The output of ${f}_{i}{ \oplus f}_{k}$ representing the correlation coefficient value is nearly 0 when one bit is changed or reversed. BIC can be verified by change in one bit value of input must satisfy both the SAC and nonlinearity.

5.4 Differential and Linear Probability

To analyse how far the S-box created can resist different attack, differential probability (DP) is calculated based on the XOR distributions of input and outputs of boolean functions. It computes the maximum likelihood on differential outputs for the given all inputs with a total of 2ⁿ elements. The smaller value of DP indicates that the S-Box generated highly resists differential attacks.

Masking of input and outputs are done for the S-box with maximum amount of linear estimates n analysing linear probability (LP). Smaller value of LP indicates that the S-Box generated highly resists linear attacks. Based on the probability equations [20], the DP and LP values are calculated for the generated S-box and its values are 0.0383 and 0.142 respectively. Thus, the generated S-box strongly resists linear and differential cryptanalysis attacks.

5.5 Mean Square Error (MSE)

MSE is calculated as the averaged sum of squared pixel values difference between the original frame as well as the resultant frame and hence the error difference of image pixels for each frames is computed. It can be expressed as

$$MSE=\frac{1}{N\times M}\sum _{n=1}^{N}\sum _{m=1}^{M}{({f}_{i}\left(n,m\right)-{f}_{o}\left(n,m\right))}^{2}$$

Where ${f}_{i}$ is the input original frame and ${f}_{o}$is the output resultant frame.

5.6 Peak Signal to Noise Ratio (PSNR)

The quality of the reconstructed image in a particular frame associated with original image is estimated by computing degree of the image based on MSE value. PSNR is estimated as

$$PSNR=10\text{log}\frac{{s}^{2}}{MSE}$$

Where s = 255 for 8 bit image. Higher PSNR value of the image in a frame indicates good quality whereas lesser PSNR value indicates low quality of the image.

5.7 Structural Similarity (SSIM) index

Based on the structural distortion measurement, the structural similarity metric is calculated among two images instead of error. The value of SSIM lies between − 1 to 1 indicating the quality degree of the visual. Higher SSIM value indicates good similarity between original and reconstructed frame. Table 1 shows the objective video quality analysis performance metrics of the proposed algorithm for particular frames of the given video.

Table 1

Performance evaluation in objective quality analysis
Frame	Entropy	PSNR (db)	MSE	SSIM
Frame 1	7.8682	10.2377	1.50	0.13
Frame 17	7.9795	9.8979	1.47	0.021
Frame 26	7.9833	9.0734	1.26	0.015

5.8 Time and Power Consumption

The existing symmetric algorithm and the proposed modified AES is compared to check the time consumption and power consumption of each while encryption and decryption. Figure 3 shows the time consumed the proposed chaos based modified AES for different video data sizes such as 250MB, 500MB, 750MB up to 1050MB.

From the performance graph, it indicates that the proposed algorithm has less time complexity for encrypting and decrypting varying size video.

Figure 4 represents the power consumed by both traditional AES and proposed chaos based modified AES for varying data sizes. The performance is analysed based on battery usage.

In this paper, selective video encryption scheme is proposed to encrypt the extracted semantic elements of the videos using HEVC encoding. A new 2D discrete hyperchaotic map based S-box is generated in the proposed framework and thus produces sequences of pseudo random bits which is used to substitute bytes dynamically during encryption of video frames. Designed S-box also performs self scrambling and good diffusion and thus has obtained high nonlinear degree, large chaotic behavior, highly complexity and more sensitive. The chaotic characteristic of S-box efficiently diminishes the correlation among pixels. The performance analysis of the modified AES with chaos based S-box has high resistance to security attacks. The experimental results shows that the proposed video encryption scheme has superior efficiency impact in less time complexity.

Availability of data and material

Not applicable.

Competing interests

The authors declare that they have no competing interests

Funding

Not applicable.

Authors' contributions

Both coauthors contributed significantly to the research and this paper, and the first author is the main contributor.

Acknowledgements

Not applicable.

Awan, I. A., Shiraz, M., Hashmi, M. U., Shaheen, Q., Akhtar, R., &Ditta, A. (2020). Secure framework enhancing AES algorithm in cloud computing. Security and Communication Networks, 2020.
Guan, B., Xu, D., & Li, Q. (2020). An efficient commutative encryption and data hiding scheme for HEVC video. IEEE Access, 8, 60232-60245.
Malik, M. S. M., Ali, M. A., Khan, M. A., Ehatisham-Ul-Haq, M., Shah, S. N. M., Rehman, M., & Ahmad, W. (2020). Generation of highly nonlinear and dynamic AES substitution-boxes (S-boxes) using chaos-based rotational matrices. IEEE Access, 8, 35682-35695.
Shah, R. A., Asghar, M. N., Abdullah, S., Kanwal, N., &Fleury, M. (2020). SLEPX: an efficient lightweight cipher for visual protection of scalable HEVC extension. IEEE Access, 8, 187784-187807.
Qiao, Z., El Assad, S., &Taralova, I. (2020). Design of secure cryptosystem based on chaotic components and AES S-Box. AEU-International Journal of Electronics and Communications, 121, 153205.
Zhu, D., Tong, X., Zhang, M., & Wang, Z. (2020). A new S-box generation method and advanced design based on combined chaotic system. Symmetry, 12(12), 2087.
Yasser, I., Mohamed, M. A., Samra, A. S., &Khalifa, F. (2020). A chaotic-based encryption/decryption framework for secure multimedia communications. Entropy, 22(11), 1253.
He, J., Xu, Y., Luo, W., Tang, S., & Huang, J. (2020). A novel selective encryption scheme for H. 264/AVC video with improved visual security. Signal Processing: Image Communication, 89, 115994.
Cheng, S., Wang, L., Ao, N., & Han, Q. (2020). A selective video encryption scheme based on coding characteristics. Symmetry, 12(3), 332.
Zhang, X., Seo, S. H., & Wang, C. (2018). A lightweight encryption method for privacy protection in surveillance videos. IEEE Access, 6, 18074-18087.
Shah, R. A., Asghar, M. N., Abdullah, S., Fleury, M., &Gohar, N. (2019). Effectiveness of crypto-transcoding for H. 264/AVC and HEVC video bit-streams. Multimedia Tools and Applications, 78(15), 21455-21484.
Hussain, I., Anees, A., Al-Maadeed, T. A., & Mustafa, M. T. (2019). Construction of s-box based on chaotic map and algebraic structures. Symmetry, 11(3), 351.
Sallam, A. I., El-Rabaie, E. S. M., &Faragallah, O. S. (2018). Efficient HEVC selective stream encryption using chaotic logistic map. Multimedia Systems, 24(4), 419-437.
Silva-García, V. M., Flores-Carapia, R., Rentería-Márquez, C., Luna-Benoso, B., &Aldape-Pérez, M. (2018). Substitution box generation using Chaos: An image encryption application. Applied Mathematics and Computation, 332, 123-135.
Hamidouche, W., Farajallah, M., Sidaty, N., El Assad, S., &Deforges, O. (2017). Real-time selective video encryption based on the chaos system in scalable HEVC extension. Signal Processing: Image Communication, 58, 73-86.
Valli, D., &Ganesan, K. (2017). Chaos based video encryption using maps and Ikeda time delay system. The European Physical Journal Plus, 132(12), 1-18.
Taha, M. A., Assad, S. E., Queudet, A., &Deforges, O. (2017). Design and efficient implementation of a chaos-based stream cipher. International Journal of Internet Technology and Secured Transactions, 7(2), 89-114.
Tian, Y., & Lu, Z. (2016). S-box: Six-dimensional compound hyperchaotic map and artificial bee colony algorithm. Journal of Systems Engineering and Electronics, 27(1), 232-241.
Namasudra, S., Chakraborty, R., Majumder, A., &Moparthi, N. R. (2020). Securing Multimedia by Using DNA-Based Encryption in the Cloud Computing Environment. ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 16(3s), 1-19.
Yang, C., Wei, X., & Wang, C. (2021). S-Box Design Based on 2D Multiple Collapse Chaotic Map and Their Application in Image Encryption. Entropy, 23(10), 1312.
Chai, X., Gan, Z., Yuan, K., Chen, Y., & Liu, X. (2019). A novel image encryption scheme based on DNA sequence operations and chaotic systems. Neural Computing and Applications, 31(1), 219-237.
Hong, S. S., & Han, M. M. (2014). The study of selective encryption of motion vector based on the S-Box for the security improvement in the process of video. Multimedia tools and applications, 71(3), 1577-1597.
Ma, X., Zeng, W. K., Yang, L. T., Zou, D., & Jin, H. (2015). Lossless ROI privacy protection of H. 264/AVC compressed surveillance videos. IEEE Transactions on Emerging Topics in Computing, 4(3), 349-362.

Download PDF

Reviews received at journal
19 Mar, 2022
Reviewers invited by journal
14 Mar, 2022
Editor assigned by journal
07 Mar, 2022
First submitted to journal
04 Mar, 2022

You are reading this latest preprint version

Selective Multimedia Data Encryption Based on Discrete Hyperchaotic S-Box Generation with Motion-Based Fragmentation

Status:

Version 1

Abstract

Figures

1. Introduction

2. Preliminaries

2.1 Fragmentation of video

2.2 Traditional Encryption Algorithm

3. Chaotic Maps And Bijective Function

4. Proposed Selective Encryption Privacy Protocol

4.1 S-Box generation

5. Experimental Results And Analysis

5.1 Nonlinearity

5.2 Strict Avalanche Criterion

5.3 Bit Independence Criterion (BIC)

5.4 Differential and Linear Probability

5.5 Mean Square Error (MSE)

5.6 Peak Signal to Noise Ratio (PSNR)

5.7 Structural Similarity (SSIM) index

5.8 Time and Power Consumption

6. Conclusion

Declarations

References

Status:

Version 1