Memetic Optimization with Cryptographic Encryption for Secure Medical Data Transmission in IoT-Based Distributed Systems

: In the healthcare system, the Internet of Things (IoT) based distributed systems play a vital role in transferring the medical-related documents and information among the organizations to reduce the replication in medical tests. This datum is sensitive, and hence security is a must in transforming the sensational contents. In this paper, an Evolutionary Algorithm, namely the Memetic Algorithm is used for encrypting the text messages. The encrypted information is then inserted into the medical images using Discrete Wavelet Transform 1 level and 2 levels. The reverse method of the Memetic Algorithm is implemented when extracting a hidden message from the encoded letter. To show its precision, equiva-lent to ﬁ ve RGB images and ﬁ ve Grayscale images are used to test the proposed algorithm. The results of the proposed algorithm were analyzed using statistical methods, and the proposed algorithm showed the importance of data transfer in healthcare systems in a stable environment. In the future, to embed the privacy-pre-serving of medical data, it can be extended with blockchain technology.


Introduction
In the recent decade, IoT acts as a backbone for the completely connected sensor devices for achieving integrated communication environments and their respective platforms both in terms of virtual as well as in real-world altogether [1] in terms of distributed systems. Recently, an organization called Health Information Exchange (HIE) Has implemented medical data transfer to speed up care of patients. The medical data has now become a regular event of everyday life across all hospitals. When it comes to the word internet, then the next word comes in the mind of people is security. Security is one of the major concerns when the internet 1. We developed an adaptive bio-inspired model for Encryption on medical data for secure transmission in IoT. 2. A DWT based Steganographic procedure has imposed to improve the mode of security.
3. An intuitive bio-inspired model developed to decrypt the encrypted data using the memetic algorithm. 4. An extensive implicated towards experimentation procedure to prove the significance of the proposed model using recent datasets (Example, DME Eyes Dataset and DICOM dataset).
The further sections of the paper organized into various sections. The literature review on securing the model for the transmission of medical data discussed in Section 2. Section 3 explains the Memetic Algorithm; the proposed model for encryption of sensitive information explained in Section 4. Section 5 deals with the experimental evaluation that includes the setup of simulations, experimental results and analysis, and the final section concludes the paper and suggests further research work.

Literature Review
In 2018, Razzaq et al. [2] composed a detailed survey on the security issues concern in IoT. Security controls on different issues such as authentication, confidentiality, are discussed in detail. In 2016, Bairagi et al. [3] presented a model namely 3 coloured image steganography techniques to provide security for efficient data transmission in IoT networks. Among the three techniques, the first and third technique makes use of red, blue and green channels to offer heightened security while the second technique takes green and blue. In 2015, Anwer et al. [4] proposed a technique and tested the model with a medical image data set to secure the images from the intruder. AES encryption algorithm was used in the generic technique to encrypt the medical images. However, the proposed model helped to achieve security in terms of integrity, authorization, and availability. In the year 2018, a detailed mobile app vulnerable on medical stream has been published, which detailed the various vulnerabilities and risk factors in the stream [5].
Razzaq et al. [6] proposed a fusion for achieving security using encryption, steganography, and watermarking methods. The proposed methodology consists of XOR operator for cover image encryption, Least Significant Bits for embedding encrypted data into the cover image, and watermarking using spatial and frequency domains. In the year 2016, Choudary et al. [8] used a decision tree model for encryption of text data of a patient inside the medical image of the respective patient. They used the steganographic technique to achieve data hiding. BFS and RSA algorithms are used for embedding the text inside the image. Zaw et al. [10] proposed a methodology for efficient transfer of medical data after being encrypted using the proposed algorithm. According to the proposed method, the given image can be divided into the standard number of blocks each about the same size. The blocks arranged for proceeding with a transformation algorithm. After the transformation process, each block encrypted using Blowfish Algorithm. Form the result analysis, and it is evident that that proposed algorithm reduces the correlation when the entropy increases. Sreekutty et al. [19] improved the security in medical image transformation using medical integrity verification system. The system consists of two-stage processes to protect and verify secret messages to achieve confidentiality and Integrity. Bashir et al. [27] proposed a new concept for encrypting medical images using AES and the shift image block integration.
In this method, the shifting algorithm used to divide the image into the number of blocks. Shifting technique is used to shuffle image rows and columns, so that shifted content defers from the original content. Muhammed et al. [28] proposed a secure method to address the RGB images via Gray Level Modification (GLM) and Multi-Level Encryption (MLE). In recent years bio-inspired methods are also used for image encryption and cryptography theory which can be found in [30,31]. Some of the other recent algorithms in bio-inspired computing applied in [18][19][20]. Prabu et al. worked on multi-discipline in the image processing stream, which includes cryptography technique and watermarking techniques [32][33][34]. Pal et al. [35] have discussed resource allocation algorithms in cloud-based distributed systems and also VM utilization, which will help the researchers to comprehend the procedure of implementing the cryptographic algorithm in IoT based distributed systems.
From the literature, it is evident that there is a need to enhance the security model for transmission of medical data and hence Memetic Algorithm can be used for effective encryption mechanisms of medical images for efficient data transmission model of text messages.

Memetic Algorithm
A Memetic Algorithm is a population-based metaheuristic algorithm that includes the standard evolutionary model as well as the local heuristic method for optimal convergence towards the optimal solution. Memetic Algorithm addresses an optimization problem with a specific procedure which is described in the following session iteratively. It aims to find the optimal solution for a problem that forms a pool of solutions called population. There are four summative methods which are defined as follows: 1. Parent Selection: The parents selected in each iteration which intends to participate in further iterations for generation quality offspring. The parents selected for the reproduction process, which generates new offspring with high quality of solutions. The high quality is indicated by the fitness of the solution. Fitness of a solution is considered as the objective function of the problem which has two modes: maximization and minimization.
2. Combination of Parents: combining two different parents to generate new offspring will lead to the generation of new solutions with better quality. Better solutions get emerged by the fusion of two quality parents.
3. Offspring Improvement Procedure: Offspring improvement deals with the heuristic procedure, which improves the quality of the offspring to the maximum extent it can be. The choice of heuristic function is highly dependent on the problem. 4. Population Update: At this point, a decision promoted on whether to engage or substitute the solution in the next generation. The decision process builts on the quality of the solution and diversity. There are two basic update rules in MA. The quality-based rule replaces the worst solution, and the diversity-based rule replaces the solution, which is like each other.

Memetic Algorithm on Secure Medical Data Transmission
In this section, a detailed model of a secure data transmission model for communicating medical images along with hidden text messages using the Memetic Algorithm. The proposed scheme consists of four subsections for effective, secure data transmission: (1) Sensitive information is encrypted using Memetic Algorithm.
(2) The encrypted information is embedded or hidden in the patient's medical image using DWT and generated a steganographic image.
(3) the information extracted from the hidden image and (4) The extracted data decrypted to get the original information. A detailed explanation of each phase given below.

Information Encryption Using Memetic Algorithm
The memetic algorithm used to address problems with NP-Hard, where the search space is unbounded and has several constraints. For this data encryption, the information encrypted to indicate this is due to intruders or hackers. AES and RSA algorithms used to encrypt the data. However, when the brute force attack applied to it, it can be easily decrypted. Hence the choice of the Memetic Algorithm through its combination with the generation of Pseudorandom numbers can be a better choice. Hence the choice of Pseudorandom number generation is applied as a local search in the memetic algorithm. In this section, the generation of the pseudorandom numbers and the image encryption method explained in detail.

Pseudorandom Number Generation
Pseudorandom numbers used to generate a random number technically. This method makes multiplicative congruential generator also called a power residue generator. The generation of the pseudorandom number is as follows: where m is a positive integer and a is a constant; Z represents the pseudorandom number. The equation means that the value of the pseudorandom number in the last iteration i will have a constant multiplication factor with a and the result divided by m. Choosing m and a have a set of rules which are defined as follows: 1. The random number generated should be less than, or equal to, m. Hence, random number collection does not exceed 2147383648. M should be picked wide enough.
2. Choosing 'a' is a prime number to m. Thus, choosing 'a' may be an odd number with a 2 16 þ 3 limit.

Memetic Algorithm On Data Encryption
The encryption schema depicts in the form of Algorithm 6.  After the successful embedding process of encrypted text to the image, the image transmits through any wired or wireless channel to the destined user. Extraction of encrypted text from the image has to be taken care of in the receiver hand. For that purpose, 2D-DWT-2L will be used to extract the encrypted information from the image file. When out of the picture, the encrypted text removed then, the cover image reconstructed using IDWT2 for both the second and first stages. Fig. 2 provides a detailed explanation of the operation. 6. Applying Crossover: for each i = 1 : |Pop|/2 do C i = Crossover(P 1 ,P 2 ,Pse); 7. Applying Mutation: for each i = 1 : |Pop| do C i = Mutation (P i ); Until (termination condition satisfied) 8

The Decryption of Confidential Information
This process refers to converting an encrypted message to its original form of text. The reverse technique applied to the encryption method. The key the sender uses to decrypt the encrypted message must be used by the recipient.  Peak Signal to Noise Ratio (PSNR): It computes the imperceptibility of the steganographic image [19]. When the PSNR is high, then it states that the steganographic image has a higher quality. PSNR calculated as Bit Error Rate ðBERÞ; PSNR ¼ 10log 10 where P denotes the maximum pixel value in an image.

Mean Square Error (MSE):
This measures the error value in terms of an average magnitude error between the original and steganographic images [20]. Mathematically represented as where N and M represent the rows and columns of the image and C ij and S ij represents the strength respectively of each pixel of the cover and the steganographic image.
Bit Error Rate (BER): It calculates the deviation of the bits that transformed due to the attenuation noise or any other noise [21]. It calculated as where E denotes the errors.
Structural Similarity (SSIM): It measures the similarity in terms of their structure between the cover and steganographic image [22]. It can be calculated as where l represents the mean and rrepresent the standard deviation.
Structural Content (SC): It measures the similarity between the cover and the steganographic image [23]. Further calculated as where C represents the cover image, and O represents the original image.
Correlation: This determines the similarity and disparity between the magnitude and the data step [24]. Further calculated as where X denotes the pairs in the information, O is the original image, and S is the steganographic image. In Fig. 3. different color and grey images used for evaluation purpose.

Security Analysis
The study of the steganographic image performs over the original image. The proposed Memetic Algorithm tested with various text sizes, and the hidden images are all images of color and white. The text messages analyzed before and during the encryption and decryption process. This study will prove that a lesser number of distortions occur before and after embedding the hidden message in the picture. The output of the proposed algorithm evaluated through two separate datasets. DME Eyes Dataset [25] and DICOM dataset [26].
For Colored images comparing the experimental results from Tab. 1. infers that as the packet size increases, the PSNR value decreases in DWT 2L and DWT 1L. On comparing the results between high packet and small packet sizes on Image 1, DWT 2L shows the improvement of 9.51% and DWT 1L with 22.75%. For Gray images on comparing the experimental results from Tab. 2. it can be inferred that as the packet size increases, the PSNR value decreases in DWT 2L and DWT 1L. On comparing the results between high packet and small packet sizes on Image 1, DWT 2L shows the improvement of 8.41% and DWT 1L with 21.70%. From Tabs. 3 and 4. it is evident that the proposed method achieves better PSNR value with less MSE when compared with the other algorithms in the existing approaches.

Conclusion and Future Work
In this paper, the Evolutionary algorithm, namely the Memetic Algorithm, is used for the secure transmission of medical images using encryption and Steganography. The results of the proposed algorithm evaluated using the performance metrics such as PSNR, MSE, SSIM, Correlation, SC, and BER. The results show the significance of the proposed algorithm over the existing methodologies. On comparing the histogram of the covered and original messages both in color and grayscale images, there is not much deviation in PSNR values which states that the proposed algorithm performs better in encryption and decryption procedure. Thus, the security concerns in healthcare systems via IoT are highly secure and confidential. The future work of this proposed method is to reduce the computational time of the encryption process.
Funding Statement: The authors received no specific funding for this study.

Conflicts of Interest:
The authors declare that they have no conflicts of interest to report regarding the present study.