Modified Advanced Encryption Standard for Boost Image Encryption

1. INTRODUCTION

The world is evolving due to the advancements in technology, attitudes, and investment. Massive amounts of data are being sent, received, and stored by government, commercial, consumer, and manufacturing groups for intelligence, safety, and other purposes such as medical, video monitoring, and IoT systems. The security requirements for images originate from the need to protect its data from two types of threats: Unintentional loss and corruption, and intentional unauthorized access or manipulation. Cybersecurity is generally recognized as the most basic and effective means of protecting images against both unintentional and intentional fraud [1]. Images data (original data) are converted into an encrypted version (encrypted information) to be saved or sent through an unsecured channel. Encrypted transmission is safe if only the sender and receiver could access the cipher and decipher algorithms and keys. The process of ensuring the integrity, availability, and privacy of data is known as information security.

Nowadays, the majority of businesses and individuals keep their data on computers. Thus, people have more access to information kept in computer databases. Much of the Stored data are very sensitive and should not be shared with the general public. Data encryption is a common technique for protecting secret information by encrypting it with a pre-existing or pre-written algorithm. The key generation is the most potent encryption aspect, which is divided into two parts: Symmetric key generation and asymmetric key generation. With the help of modern high-performance computing devices, hackers may quickly break the key. Therefore, encrypted data that cryptographic analyzers cannot decrypt are currently in demand [2]. The encryption and decryption processes are affected by the encrypted file size and the length of the key [3]. The key used for encryption/decryption in symmetric cryptography is the same. As a result, the required distribution must occur before information transfer. In symmetric key cryptography, as shown in Fig. 1, the key is highly significant because the system’s security is directly related to the key’s characteristics, such as key length. Advanced Encryption Standard (AES), DES, TRIPLE DES, BLOWFISH, RC4, and RC6 are examples of symmetric cryptography [4].

Fig. 1. Symmetric key cryptography.

AES is one of the most used symmetric algorithms and a new generation of data encryption standards. It offers the most outstanding safety and highest operation speed [5]. The AES is the next-generation cryptographic system with many benefits, including strict security, high efficiency, customization, and convenience.

The research establishment has shown keen interest in image encryption to protect valuable images from hackers in recent years. As a result, several encryption algorithms have been used for encrypting images to achieve this goal. This research aims to show the efficiency and speed of the modified AES algorithm compared to the original AES.

2. LITERATURE REVIEW

With the rapid growth of digital data interchange, secure data transfer is becoming increasingly crucial. since images are increasingly being used in different processes, it is vital to keep confidential image data safe from unauthorized access [6]. In Maolood et al. [7], modifications were made to the original AES algorithm to increase security. Using evaluation measures, the effectiveness of the modifications was demonstrated, and the improved algorithm was evaluated. The results showed a small increase in the encryption time of the modified AES algorithm compared to the original AES algorithm because of adding several layers of security. The authors in Ahamad and Abdullah [8], compared the four encryption algorithms (Blowfish, AES, XOR, and RSA), applied to four types of data represented by: Text, image, audio, and video. The simulations demonstrate that AES is more time efficient than all other methods. In another work [9], the AES round numbers have been increased to 16 to be more secure. Theoretical study and practical results revealed that this AES strategy provides great speeds and less data transmission over unprotected networks. Moreover, the research study in Arman et al. [10] focuses on symmetric Key cryptosystem represented by the AES algorithm and proposes a significant improvement over AES in terms of privacy and ease of operation. The goal was to achieve a rapid implementation time and low memory consumption while maintaining a high security level. As a result, this solution obtained 88% of efficiency and a faster net processing time. This is a big change in terms of efficiency over the standard AES. In the more advanced researches, to address the poor diffusion rate in the early rounds, the AES cipher round and key schedule have been updated to include more primitive operations such exclusive OR and modulo operations [11]. According to the avalanche effect evaluation results, the results show that the modified AES has enhanced propagation and ambiguity features because there was a 61.98% increase in dispersal in round 1, 14.79% in round 2, and 13.87% in round 3. However, in Shakor and Surameery [12], AES is combined with elliptic curve cryptography (ECC) algorithm which has been used to protect the sensitive COVID-19 files that are stored in the cloud. Although different layers of security have been added, the results indicated that the impact of the hybrid system was more significant than the other secure algorithms without affecting the consumption time. The author in Mohd and Ashawesh [13] proposed a secured modified AES (MAES) algorithm, which reduces the rounds of AES to 14 rounds to decrees the consumption time of the encryption and decryption process, and improve the digital data security. The results prove that, compared with other studies, the proposed method uses less encryption and decryption time while providing better security, as confirmed by the avalanche effect test. Moreover, in Lin et al. [14], an improved AES cryptosystem based on chaotic synchronization with dynamic random keys is proposed. In the traditional AES encryption system, static keys are used, which must be exchanged in advance and confirmed for safe preservation. The chaotic system’s synchronization technology, on the other hand, was used to overcome the inadequacies of key storage. In the proposed approach, the static key becomes dynamic and random, and it no longer needs to be kept or transmitted over an open channel. In Hafsa et al. [15], the authors use an efficient type of a built-in approach that utilized the AES and elliptic curve encryption (ECC) algorithms for encrypting medical images. It combines the advantages of symmetric AES accelerated data encryption and asymmetric ECC to ensure the exchange of symmetric session keys. The security analysis was successfully performed, and their experiments proved that the proposed technology provides a more straightforward and correct cryptographic basis. Furthermore, the evaluation results demonstrate the proposed algorithm’s effectiveness, rapidity, and high security.

Hence, after studying various previous researches on the modification of AES, we came up with a new technique that increases the processing speed and data security.

3. AES

The AES is a symmetric key method that has been developed by the United States National Institute of Standards and Technology as the standard for encrypting digital data. AES is the most powerful algorithm in recent times since it is the only vulnerable to brute force attacks, which makes it difficult for cryptanalysts to crack. It is commonly used in banks and organizations to protect critical information because it promises cybersecurity [16]. The size of the key determines the number of rounds for encryption and decryption in AES, which could be 10, 12, or 14 rounds for 128, 192, and 256 bit keys, respectively [17]. AES is one of the most widely used symmetric key algorithms and a new generation of data encryption standards. It offers the greatest secrecy and highest operation speed [5].

As shown in Fig. 2, the AES algorithm comprises four invertible transformations: SubBytes, ShiftRows, MixColumns, and AddRoundKey. All of the encryption rounds use these transforms, with the exception of the end round, which removes the MixColumns transformation to make the encryption and decryption methods symmetric.

Fig. 2. Structure of AES algorithm.

The encryption process consists of a number of different transformations applied consecutively over the data block bits, in a fixed number of iterations, called rounds. The number of rounds depends on the length of the key used for the encryption process. The AES encryption process is as follows

Cipher(byte in[X*Nb], byte out [X*Nb], word w[Nv*(Nr+1)])

Begin

Byte state[X, Nb]

State=in

AddRoundKey(state, w[0,Nb-1])

For round=1 step 1 to Nr-1

SubBytes(state)

ShifRows(state)

MixColumns(state)

AddRounKey(state, w[round*Nb, (round+1)*Nb-1])

end for

SubBytes(state)

ShifRows(state)

AddRounKey(state, w[Nr*Nb, (Nr+1)*Nb-1])

Out=state

End

4. MAES ALGORITHM

To secure the image files and increase their privacy, a modified AES algorithm is employed to encrypt them. The AES is modified to address its higher computational requirement due to the complex mathematical operations in MixColumns transformation making the encryption process slow. Bit permutation includes merely shifting the positions of bits in each stage and does not require any significant mathematical computation. The bit permutation transformation takes the place of the MixColumns transformation in AES, Fig. 3, depicts the bit permutation modification of AES.

Fig. 3. Structure of MAES algorithm.

For the encryption process, take the state value per column (for example, column 0) from ShiftRows transformation. Each column has four rows, as shown in Fig. 4. Moreover, each state in column 0 is composed of eight bits, represented as ((x,0), b) where x, 0, and b indicate the row, column, and bit number for each state, respectively. Subsequently, the states are partitioned into four blocks matrix, called block 0, block 1, block 2, and block 3 and the transpose of each block matrix is the next step of the process. Finally, to get the new value at state a’(x,y), we need to do a row-wise concatenation of the bit values of the transposed block where x is the column_value and y is the block_value. MAES encryption process is as follows:

Cipher(byte in[X*Nb], byte out [X*Nb], word w[Nv*(Nr+1)])

Begin

Byte state[X, Nb]

State=in

AddRoundKey(state, w[0,Nb-1])

For round=1 step 1 to Nr-1

SubBytes(state)

ShifRows(state)

Bit Permutation(state)

AddRounKey(state, w[round*Nb, (round+1)*Nb-1])

end for

SubBytes(state)

ShifRows(state)

AddRounKey(state, w[Nr*Nb, (Nr+1)*Nb-1])

Out=state

End

Fig. 4. Bit permutation.

For the decryption process, the inverse bit permutation is utilized. There are four columns in each row. In the first steps, starting with row 0, we take the state value from the InvSubBytes transformation. Next, each state is made up of 8 bits, denoted by ((0, y), b), where 0 denotes row 0, y denotes column, and b denotes the bit position in each state. Taking the number of bits per state creates 1×32 matrix. Subsequently, the matrix is partitioned into eight blocks, called block 0 through block 7. After that, the blocks 0 to 7 are transposed, yielding eight 4×1 transposed matrices. Finally, to get the new value at state a’(x,y), we need to do row-wise concatenation of the bit values of the transposed block, where x is the block_value and y is the row_value. Fig. 5.

Fig. 5. Inverse bit permutation.

5. EXPERIMENTAL RESULTS AND DISCUSSION

This section shows the experiment run to derive the effect of image size on the encryption and decryption time and includes a comparison between the modified AES system and the existing standard AES solution. To apply security, AES-128, AES-256, MAES-128, and MAES-256 algorithms were used in the encryption and decryption processes.

These two processes were implemented using the Python programming language. Both standard and modified AES algorithms are written in Spyder, the Scientific Python Development Environment, and simulated on Intel Core i7-2.8 GHz processor with 16 GB RAM and running over Mac OS High Sierra.

The main purpose of this study is to calculate the encryption and decryption speed of each of the algorithm under the study of different image sizes. The AES and MAES algorithms were evaluated for 10 trials to determine the average encryption and decryption time. Their implementation is tried to optimize the maximum performance for the algorithm. Therefore, the throughput for encryption is calculated for each algorithm. Encryption time is used to calculate the throughput of an encryption scheme. The throughput of the encryption scheme is calculated by dividing the total plain image size in kilobytes by total encryption time in second for each algorithm. If the throughput value is increased, the power consumption of this encryption technique is decreased.

Throughput = Plaintext (KB)/Encryption time (Sec.)

In this research, different sizes of image files were experimented such as 200 KB, 400 KB, 800 KB, and 1600 KB. The performance metrics are analyzed as encryption and decryption time and throughput. The performance of the encryption and decryption time was measured in milliseconds.

For the 128 bit key, it was observed that both the encryption and decryption times increase as the image size increases, as shown in Tables 1 and 2. However, it was noticed that the encryption and decryption processes of the MAES take less time than the standard one for the same image size.

TABLE 1: Comparison of AES and MAES encryption time using 128 bit key size

TABLE 2: Comparison of AES and MAES decryption time using 128 bit key size

Regarding the enciphering process, for the file sizes of 200, 400, 800, and 1600 KB, the encryption time of the AES was 28.24, 52.11, 113.11, and 242.89 ms, respectively. In comparison, the encryption time of the MAES for the same file sizes were 27.22, 51.26, 111.45, and 240.04 ms, respectively. Moreover, the decryption time of the AES was 28.24, 52.11, 113.11, and 242.89 ms, respectively. In contrast, the decryption time of the MAES for the same file sizes were 27.22, 51.26, 111.45, and 240.04 ms, respectively.

On the other hand, ultra-high security may be achieved using AES-256 due to the number of rounds, which provides privacy to unauthorized users. Therefore, the system has been tested with AES-256 and MAES-256 for different file sizes. For the file sizes of 200, 400, 800, and 1600 KB, the encryption process of MAES-256 was faster than AES-256 by 0.74, 1.91, 7.37, and 6.91 ms, respectively. In the same way, the decryption process of MAES-256 was faster than AES-256 by 3.94, 3.00, 6.82, and 11.0991 ms, respectively, as shown in Tables 3 and 4. Thus, in comparison with AES, the presented results show that slightly less encryption and decryption times in milliseconds were recorded with the bit permutation in MAES for all of the image file sizes, which increases the security, overall performance, and effectiveness of the system.

TABLE 3: Comparison of AES and MAES encryption time using 256 bit key size

TABLE 4: Comparison of AES and MAES decryption time using 256 bit key size

By considering the results in Tables 1-4, the average execution time for image data types of AES are 436.35, 450.11, 455.27, and 472.2, respectively, and MAES are 429.97, 444.13, 441.35, and 466.2, respectively. Similarly, the throughput are (6.87, 6.6650, 6.589, and 6.353) and (6.97, 6.754779, 6.797, and 6.435) for AES and MAES, respectively. The results show that the standard AES algorithm has higher encryption average time and the MAES has higher throughput. Hence, MAES is more efficient in image data encryption than AES.

Besides, a comparative analysis of the obtained results was carried out with an existing work existing works. This was a little bit of tasking, as there are no standard performance metrics that are widely and generally acceptable by all researchers in this regard. While some measured the performance of their modified AES version using text files of different sizes, some used images and video files. However, most authors employed execution time as their performance metrics and just few researches employed the encryption and decryption times, separately.

Therefore, the proposed technique was compared with the recent solutions for the “Implementation of Modified AES as Image Encryption Scheme” that was proposed in 2018 [18]. The research was available to investigate the use of modified AES for image encryption.

As presented in Table 5, the proposed technique provides optimistic solutions by recording lower execution time in milliseconds than other research solution. Although this may not be clearly noticeable in real-life application, yet it is significant.

TABLE 5: Comparison of proposed method with another research solutions

6. CONCLUSION

Cryptography is a widely used approach for ensuring data transfer and image storage security. The proposed algorithm aims to reinforce and improve the standard AES method by replacing the MixColumns transformation with bit permutation. With different file sizes, the new algorithm is tested for encryption/decryption time in milliseconds. The results reveal that the MAES achieves a higher level of security with requiring less time. The key sizes used to compare AES and MAES were 128 bits and 256 bits.

The results showed that the encrypting time in the files whose size was between 200 and 1600 KB in the MAES algorithm was less than AES by 1.595% with the128 bit key size, while MAES has a lower percentage of encryption time than AES by 3.48% when applying a 256 bit key size. In the same way, MAES decryption time was 1.495% faster than AES with a 128 bit key size and 1.5% faster with a 256 bit key size. In addition, the results show that the standard AES algorithm has higher encryption average time and the MAES has higher throughput. Hence, MAES is more efficient in image data encryption than AES. Furthermore, the proposed MAES cryptosystem provides improved security in terms of protection against different types of attacks. As a part of future work, we can use additional linear processes experiments to boost AESs confusion property.