Robust Cyber-Physical System Enabled Smart Healthcare Unit Using Blockchain Technology

Abstract

With the growing demand for smart, secure, and intelligent solutions, Industry 4.0 has emerged as the future of various applications. One of the primary sectors that are becoming more vulnerable to security assaults like ransomware is the healthcare sector. Researchers have proposed various mechanisms in smart and secure health care systems with this vision in mind. Existing systems are vulnerable to security attacks on medical data. It is required to build a real-time diagnosis device using a cyber-physical system with blockchain technology in a considerable manner. The proposed work’s main purpose is to build secure, real-time preservation and tamper-proof control of medical data. In this work, the Bayesian grey filter-based convolution neural network (BGF-CNN) approach is used to enhance accuracy and reduce time complexity and overhead. Additionally, PSO and GWO optimization techniques are used to improve network performance. As an outcome of the proposed work, the privacy preservation of medical data is improved with a high accuracy rate by a blockchain-based cyber-physical system using a deep neural network (BGF Blockchain). To summarize, the proposed system helps in the privacy preservation of medical data along with a reduction in communication overhead using the Bayesian Grey Filter–CNN.

1. Introduction

Industry 4.0 evolution is currently playing a vital role and demanding secure and smart applications. Advanced technologies such as blockchain technology, cybersecurity, data science, and the Internet of Things (IoT) are being used to build a system that is being used to achieve Industry 4.0 [1]. A new era has started for the healthcare sector with industry 4.0, providing a better experience for both the patients and the doctors in the future. A significant impact is being shown in the health sector through these kinds of applications.

The main objective of the proposed model is to build a secure, real-time cyber-physical healthcare system that enables an intelligent healthcare system using blockchain technology. Remote patients are concerned about securely transmitting continuous monitoring data for making significant decisions via unlimited networking capabilities [2]. The major problem with the heterogeneous devices enabled with the sensors is connectivity among the communicating nodes, such as patients, hospitals, and doctors while treating patients from remote locations. The cyber-physical system is used to enable real-time health care information services in India. The proposed cyber-physical system can resist data attacks like ransomware attacks such as Wannacry, tampering with medical data, Man in the Middle attacks, etc. This system can improve the privacy and preservation of medical data by storing it in the blockchain and using advanced mechanisms like convolution neural networks for analyzing the data.

With the features of decentralization, distributed data storage, point-to-point transmission, and encryption algorithms, blockchain technology has revealed new insight into the security and assurance of clinical records. It can resolve the logical inconsistency between information sharing and security insurance with legitimate security strategies. Intelligent healthcare systems are facing huge challenges for data security, and it has become critical to ensure the privacy of patients’ data.

These clinics are currently focusing on the most effective way to make use of practical web clinical benefits to prevent attacks on the clinical data framework and theft of individual data. As blockchain technology and the framework for cyber-physical healthcare services develop, related strategies—such as those involving clinical data, clinical record security, adaptable medical services, clinical internet business, etc.—have quickly advanced clinical and health services. Using blockchain technology, the members are connected commonly. From a comprehensive perspective, blockchain innovation should incorporate four intrinsic points: highlight point network plan, innovative encryption algorithms, usage of distributed algorithms, and distributed information storage. Others might include distributed data storage, AI, VR, the Internet of things, big data, and so on. From a narrow viewpoint, blockchain provides information storage, data set, record activity, and so on. With the advancement of technology, enormous clinical information of an individual can be uploaded to the cloud to achieve a reduction in project expenses.

Some clinical foundations scramble to encipher the data in advance and jumble it for the protection and security of patient information. Moreover, they need other clinical establishments to share certain cipher texts frequently and forestall anybody from decoding these cipher texts. Decentralization and tamper-proof computing are combined in blockchain technology. Healthcare organizations can ensure confidentiality of patient information at a cheaper cost by using blockchain hubs with distributed computing which also ensures collaboration amongst clinical associations. These will have a huge impact on the improvement of future medical services to a greater extent.

The remaining part of this work is illustrated as follows. The related work is presented in Section 2 with a summary table. Section 3 presents the proposed methodology. A detailed analysis of the results is discussed in Section 4; it mainly deals with the comparison of the present methodology with existing approaches. Section 5 consists of the conclusion and future work.

2. Materials and Methods

Wanglin Yan et al. [3] presented some details on geospatial spaces angle-based applications using IoT and cyber-physical systems (CPS). The authors have addressed advanced technologies such as the Internet of things (IoT) and Geographical information systems (GIS), which are combined to make a cyber-physical system (CPS) in urban areas.

Mobile electrocardiogram systems (ECGs) are now crucial tools for treating heart issues that affect the elderly. WenfengYin, et al. [4] proposed an application that utilizes radio ultra-wideband radar sensors to categorize the ECG data. The authors have also included a cascaded convolution neural network (CCNN) to classify the ECG signals. Gael Loubet, et al., [5] proposed a health monitoring cyber-physical system. The drawback of this work is the lack of focus on the integrity and safety of data. For addressing data integrity and security, Jian Xu, et al. [6] proposed an authenticated data structure utilizing privacy-preserving architecture. The above work is lacking in the fields of security and integrity while processing and preserving the data. It also lacks faster communication due to the non-involvement of advanced feature mechanisms.

Zhang et al., [7] examined various use cases consisting of blockchain technology for securing healthcare services. In this paper, the authors have specified the efficiency and importance of blockchain technology in designing health care systems. S. Sciancalepore et al. [8] developed a framework based on the attribute specifications of the systems. The system operates on a cloud assistance-based cyber-physical system (CPS). The proposed system meets all the requirements like data integrity, security, and reliability but lacks in maintaining huge amounts of healthcare data.

George Drosatos et al. [9] addressed the role of blockchain technology in the biomedical domain. The proposed work highlights various pieces of research conducted on blockchain technology which is considered its main strength. Here, the authors have discussed information sources, required protocols, data charting, and the results but have not provided any specifications for implementing the application. Shrier, et al. [10] discussed the infrastructure of blockchain technology in a way similar to other proposed methodologies. HaoGuo et al. [11] illustrated a technique to store health records securely with a signature-based authentication. It also comprises ABMS and ABE schemes, which provide confidentiality to the HER details. The main drawback of this work is the lack of synchronization between the system and electronic health records. It also failed to describe the analysis of implementation results.

Suat Mercan, et al. [12] illustrated a framework, which could collect the data from various Internet of Things (IoT) components and also validate the data authenticity. This system has two modules with the combination of a blockchain system. The model used the public blockchain and mentioned that the data block size was reduced by the hash functions used. The main limitation of the work is stating low process time and overhead for the proposed model without showing the comparison results.

Prabal Verma, et al. [13] discussed an approach that is based on IoT, Fog and Cloud system applications. It is referred to as a cyber-physical system. The data classification and UC diagnosis were efficiently performed using a deep neural network and Naïve Bayes classifier. An altered message will be generated from the system whenever the patient requires emergency treatment. Bahar Farahani, et al. [14] developed an Artificial Intelligence-driven IoT system. That is a collaborative machine learning approach. This work mainly focused on energy and latency factors. To address the work of S. Sciancalepore et al. [8], Amir M. Rahmani, et al. [15] designed an IoT-based system to measure a warning score as soon as possible. It has been designed as an intelligent system reliably and efficiently.

Asma Khatoon, et al. [16] designed a blockchain technology model to manage medical data with clinical and surgery trial procedures. In this work, the author discussed how data has been exchanged among the stakeholders with transparency while collecting medical data. Siyal et al. [17] examined the importance of smart contracts and blockchain technology in healthcare systems and are also concerned about how they can reduce fabrication and loss of medical data by using distributed ledger-based blockchain technology.

Daisuke, et al. [18] showed a methodology for preserving the medical records using a hyper ledger blockchain that is a permissioned blockchain. The authors have collected medical records using a smartphone. Those records have been secured on top of the private Blockchain. The main drawback of this work is that it is only designed and does not show any implementation results. Moreover, no preprocessing operations are performed on the data before going to maintenance in the Blockchain.

Rajesh Gupta, et al. [19] reviewed CPS-based systems security in terms of privacy and their protection with advanced technologies such as Artificial Intelligence (AI), Blockchain technology, and the Internet of Things (IoT). This work also addressed the role of these techniques in Industry 4.0. The main drawback of this work is unable to present implementation results. The work only specified the information about the challenges, open issues and solutions that were proposed.

P. Bhattacharya. et al. [20] discussed a method to secure medical data of CT scan images. The authors have used a recurrent convolution neural network (RCNN) to identify the region of interest (ROI) in this work. The main drawback of this work is failing to show the deployment results over a blockchain. The work mainly focused on applying deep learning techniques to the data.

Mahendra Kumar, et al. [21] proposed a hyper ledger fabric-based medical data sharing system. In this system, the authors have designed using 20 permission nodes to maintain the COVID-19 patient data. The main limitation of this work is medical data is directly maintained in the Blockchain. There is no preprocessing to remove the noise from the information which may cause it to take more time for processing and increase space complexity.

Radanliev, P, et al. [22] proposed a framework for secure COVID-19 vaccine supply chain infrastructures. Before integrating new IoT technologies, this paper proposes that supply chains must be conceptualized with ethical awareness of the cyber risks and with a comprehensive understanding of the operational and digital capabilities of each supply chain component. The effects of shared risk from coupled and complicated IoT systems are not discussed in the work.

Yan Zhuang, et al. [23] came up with a secure blockchain model for patient-centric HIE (Health Information Exchange). Using blockchain features, the patient’s health data can be stored securely and the data immutability feature ensures reliability. The data can be restricted from unauthorized access using a secure peer-to-peer network. The architecture consists of two modules: the linkage module and the request module. The linkage module is used by the administrator of each healthcare center for inputting patients’ data. The request module is used by the patients to provide access to the clinicians to access their data. The drawbacks of the model are requiring setup at each center and scalability issues.

Petar Radanliev, et al. [24] proposed a methodology for shared responsibility and ethics in health policy using Internet of Things (IoT) devices in healthcare systems. In this work, the authors have presented various bibliometric analyses on healthcare systems before the emergence of COVID-19 and the risks involved in integrating ethics into IoT-enabled medical devices. The work has not shed light on the implementation process for enhancing security; instead, it just gives a detailed analysis of the risks associated with IoT-enabled medical devices.

Anil Islam, et al. [25] proposed an approach to a federated learning-based blockchain embedded data accumulation scheme using drones for the Internet of Things. The work mainly focuses on securing remote regions using drones and blockchain technology. The work contains a two-phase authentication mechanism in which request validation is performed using a cuckoo filter followed by timestamp nonce. Thereafter, the model is preserved using blockchain. However, the proposed system is not implemented at full scale, which is the main drawback.

Table 1. Summary of Literature Survey.

Author	Limitations			Preservation Mode	Design/ Implement	BCT Type	Comp. Type	Comp. Approach
	Is CPS	Is BCT	Is Computational
[1]	No	No	Yes	GIS	Implement	Public	Machine Learning	Linear
[2]	No	No	Yes	Cloud	Design	Na	Machine Learning	ABMS
[3]	Yes	No	Yes	Fog	Design	Na	Deep Learning	CCNN
[4]	No	Yes	No	Blockchain	Design	Hyper ledger	Not used	Not used
[5]	Yes	No	Yes	Cloud	Implement	Private	Deep Learning	RCNN
[6]	Yes	No	Yes	Cloud	Implement	Na	AI	IoT
[7]	Yes	Yes	No	Blockchain	Implement	Na	Not used	Not used
[8]	No	No	No	Database	Implement	MongoDB	Not used	Not used
[9]	Yes	Yes	Yes	Blockchain	Design	Na	AI	IoT
Proposed Method	Yes	Yes	Yes	Blockchain	Implement	Ethereum	Deep Learning	BGF-Blockchain

shows the summary of the related work on the proposed system.

3. Proposed Methodology

The proposed project includes two workspaces: one is a physical space connected to medical facilities and patients, and the other is cyberspace related to blockchain technology. As a result, this system is known as a cyber-physical system. As depicted in Figure 1 cyberspace considers only physical space’s inputs for measuring the performance with the help of technical aspects.

Huge amounts of data are collected from the IoT-based sensors in the physical space. The data gives an outline of patient health information such as heartbeat, sugar, blood pressure, etc. Furthermore, a unique ID is allotted to each patient for future communication purposes which ensures endless network capabilities. Physical space contains three types of users: healthcare centers, doctors, and patients. In cyberspace, we need to operate three technical modules. Module 1 deals with the collection and analysis of patient medical data. Module 2 helps in processing the medical data based on specific constraints to get the final results. Module 3 helps in the secure preservation of the results obtained in a blockchain. The main novelty lies in the optimization of BGF-CNN parameters using the hybrid GWO-PSO technique, which helps in the classification of medical data and later securing them with the help of Ethereum-based public blockchain technology.

Figure 1 describes the architecture of the proposed methodology. Figure 1 consists of two spaces: physical and cyberspace. In physical space, medical data classification is performed using BGF-CNN and later on optimized using the GWO-PSO technique. In cyberspace, the data is secured using Ethereum-based blockchain technology.

Module-Wise Implementation Procedure

Module 1: Patient Medical Data Collection, Analysis, and Process

Medical data analysis is performed using Bayesian Grey Filter-based Convolution Neural Network (BGF-CNN) and Volume-aligned CNN.

• The main motive of the module is to collect the raw medical data from the MEHEALTH repository. This data consists of various patients’ heart signal rates.
• This data is processed by utilizing the Grey filter approach. The BGF-CNN approach is used for multiple purposes such as noise removal and classification. Volume-aligned CNN is applied to the proposed methodology to classify the results as healthy and unhealthy heart signals.

Layer 1: Input Data

In this module, the accumulated medical data collected from the data set of MEHEALTH can be extracted from the machine learning repository. This data is acquired from various IoT sensors connected to the patient body that is used to measure the patient’s heartbeat. IoT-driven sensors are used to monitor the heartbeat by choosing the ‘chest’, ‘right wrist’, and ‘left ankle’. The main stages for analyzing the medical data are Input data (Reconnaissance), preprocessing, classification, pooling, straightening the result samples, and final output.

Layer 2: Processing the data

In this module, the collected data was analyzed by incorporating IoT-based solutions and data science approaches such as Deep Neural Network-Based Gray Filter Bayesian Convulsion Neural Network (BGF-CNN) over Cyberspace. It helps to classify the medical data like healthy and unhealthy heart signals. The volume-Aligned Softmax CNN technique is used to enhance the accuracy rate of medical data analysis, and the Max Pooling function is used to classify the healthy and unhealthy heart signals. Various operations are performed in the module as shown in Figure 2. Preprocessing the collected medical data using a Gray filter is shown in the module. It is implemented using a CloudSim simulator.

As shown in the pseudo-code in Algorithm 1, initially, positive time sequence data (i.e., non-negative data) of the patient’s heart monitoring is maintained as a set i.e., Pj = {P1, P2,… Pn} where ‘j ‘is the patient source sequence number. Assembled sequence (P j + 1) generated based on the source sequence (Pj) using AGO operations to flatten the randomness. This can be achieved by using the following notion i.e., P j + 1 = {P j + 1 (1), P j + 1 (2)… P j + 1 (n)} Where P j + 1 (n) =∑ni = 1 Pj (i). Next, it generates the feature map sequence for all the patients using the average value of two consecutive patient data activities. Finally, it summarizes the detected features in the input data i.e., ‘PDi’.

Algorithm 1: Preprocess: Grey Filter pseudo code

Input: Patient Data ‘Pi’

Output: Preprocessed Data ‘PDi’

Process:

Repeat

For each patient, ‘Pi’

Express positive time sequence data as

Pj ={P1, P2,… Pn} where ‘j ‘is the patient source sequence number.

Find assembled sequence by using ‘AGO’ where AGO = assembled Multiplying operation

P j + 1 = { P j + 1 (1), P j + 1 (2)… P j + 1 (n)}

Where P j + 1 (n) =∑ni = 1 Pj (i)

Generate feature map (F). It is obtained using the average value of two consecutive activities.

Fj = { f j (2), f j (3),… f j (n),}

Where f j (n) = ½ f j (n) + ½ f j (n−1)

End for

Until (obtained feature map for all patient activities)

Return (Preprocessed Data ‘PDi’)

Layer 3: Optimization of CNN parameters (GWO-PSO)

After preprocessing all the required parameters such as learning rate, population size, hidden units, mini-batch size, and momentum, they are optimized using a hybrid GWO-PSO technique. This hybrid Grey Wolf Optimization–Particle Swarm Optimization is a meta-heuristic optimization which is used to increase network performance. It helps in reducing the tedious job of optimizing the parameters manually. It performs in a way starting with the initialization of GWO and PSO parameters followed by defining the initial cost function. Later, the fitness function is calculated, followed by calling the routine for GWO-PSO and finally updating the positions. GWO-PSO parameters give the best fitness value which improves the model agility.

Layer 4: Logistic Sigmoid Activation

Layer 3 output will become an input to the classification stage i.e., Preprocessed Data’ PD’. The bayesian binary classification function is used to identify the patient records obtained from the sensing values based on the sampling rate. Here, we considered the sample rate as ‘50 HZ’. If the sample rate is equal to the threshold value, then the patient activity will be captured or else it will not be captured. All the captured activities are maintained in a set named ‘OP’ as shown in Algorithm 2.

Algorithm 2: Classification

Input: Pre-processed Data ‘PDi’

Output: Optimal patient data ‘OP’

Process:

Initialize Sample Rate ‘SR’

For each Pre processed Data ‘PDi’

Find posterior probability using

Prob(q = 1/p) = {prob(p/q = 1)prob(q = 1)}/Prob (p) || SR

Where Prob(p) = {{prob(p/q = 1)prob(q = 1)} + {prob(p/q = 0)prob(q = 0)}}

Evaluate the likelihood ratio by

Prob(q = 1/p) = {prob(p/q = 1)/Prob (p/q =1)+ prob(p/q = 0)}|| SR

If ‘OP’ = Threshold value ‘

Human activity captured

Else

Not captured

End if

Return (Human activity captured set ‘OP’)

End for

Layer 5: Pooling

The output of the classification module i.e., ‘OP’ becomes an input to the pooling layer, which is the hidden layer. Volume Align softmax CNN is applied to the data to obtain efficient and accurate heart signal classification. The simulation of the GFB-CNN will be performed on different activities of the patients as listed in

Table 2. Activities set.

Activity	Stand	SIT & RELAX	Lying	Walk	Climb	Run	Bend	Jump
Label	A1	A2	A3	A4	A5	A6	A7	A8

.

The MEHEALTH dataset has body motion vital signals which are recorded from different volunteers while performing various tasks. Sensors were placed on the body of the patients and data was recorded at a sample rate of ‘50 Hz’.

Module 2: Blockchain: BGF_Blockchain

Blockchain-based secure cyberspace is utilized for preserving medical data using Ethereum-based public blockchain [26,27], TEST RPC, Metamask Wallet for ETH (cryptocurrency), and DAPP, that differ with the other mobile applications [28]. Ethereum is used instead of the general blockchain in the proposed model because Ethereum does not involve third parties to perform transactions and downtime is significantly low. The final step is to safely store the results in a public blockchain environment powered by Ethereum as shown in Algorithm 3. To implement this a blockchain-based decentralized application (DAPP) is designed using Test RPC and Metamask Wallet, for ETH balance (cryptocurrency). The following process illustrates how transactions using blockchain technology work:

• A decentralized application (DAPP) is used to send medical data into the blockchain.
• Configure the TEST RPC which acts like a blockchain emulator.
• Configure the metamask wallet as a chrome extension, to access the Ethereum-based blockchain-enabled DAPP.
• Generate smart contracts on TESTRPC to perform the operations over the blockchain.
• Each medical record contains transactions performed using the blockchain. Later the process on Metamask is started. ETH balance in the metamask will be deducted to operate on a block in the blockchain.
• A unique ID is allotted for each successful transaction.

Algorithm 3: Data process over a blockchain

Input: Health and unhealthy Heart Signals Data

Output: a Unique blockchain-based ID

Process

Generate a smart contract to penetrate the medical data into DAPP.

For all the patient’s medical data

if (ETH balance >= threshold balance)

Compiled and deployed the process on TESTRPC

The medical data is stored in a block as a transaction.

A unique ID is allotted to the transaction.

Else

medical data on TEST RPC is not compiled

Figure 2 demonstrates the medical data analysis using CNN as shown below.

Solidity programming and scripting language are used to design a decentralized application on TEST RPC, a blockchain emulator. To deploy the operations on Blockchain, a smart contract made of lines of code must be written using the solidity programming language. By verifying the ETH balance deduction in the Metamask wallet, the medical data smart contract must first be registered, then it must be compiled, then it must be deployed on an Ethereum-based blockchain. Any transaction on the blockchain requires payment with cryptocurrency in the form of an ETH balance [29,30,31].

ETH balance is required to compile and deploy the system operations that are maintained as smart contracts using solidity programming. Each patient’s data is stored in the Blockchain as a transaction in a specific block. All the blocks have been interconnected with the crypto hash value that depends on earlier block hash value and current block data transactions. A unique ID is allotted for all the patient’s transactions. The patient data Is tamper-proof because of blockchain technology’s advanced features such as immutability, distributive, transparency, reliability, and security [32,33].

As shown in Figure 3, Before going processing the medical data over a blockchain, the process verifies the wallet balance for Eth. The medical data is processed and maintained as a transaction in a specific block over a blockchain when the wallet has enough Eth balance, as shown in Figure 3. Without enough Eth balance, the data is not maintained in the block as a transaction.

4. Results and Analysis

The proposed simulation performance was evaluated by comparing with the existing approaches called GFB-CNN [34], adaptive IoT-enabled CPS [1], and IoMT-PLM (Internet of Medical Things with product Lifecycle management [2]). The attack rate on data is reduced because of advanced features of blockchain technology such as immutability, transparency, distribution, and security. The proposed model provides universal applicability of a secure real-time cyber-physical health care system for authorized end users. Any authorized user can use the application from anywhere and at any time by using the allotted unique BCT_ID to the patient data.

The blockchain ‘${\mathit{B}}_{\mathit{c}}$’ size depends on various factors such as instant time ‘t’ and the system’s workload. The system’s workload depends on the gas limit of a block ‘G’, the creation time of a block ‘T’, transactions ‘$\mathit{t}\mathit{x}\mathit{n}$’ included time in a block ‘${\mathit{B}}_{\mathit{t}\mathit{x}\mathit{n}}$’ and header size of block ‘H’.

The total time of transactions in a block is evaluated approximately by using the following Equation (1).

$${\mathit{B}}_{\mathit{t}\mathit{x}\mathit{n}}={\sum }_{\mathit{i}=\mathbf{0}}^{\mathit{i}=\mathit{n}-\mathbf{1}}\mathit{t}\mathit{x}{\mathit{n}}_{\mathit{i}}$$

(1)

The header of block size is ‘${\mathit{H}}_{\mathit{s}}$’, the rate of block creation is constant i.e., ‘C’. Furthermore, the header size of a block and its creation time is considered to evaluate the block creation time as shown in Equation (2):

$${\mathit{B}}_{\mathit{t}\mathit{i}\mathit{m}\mathit{e}}={\displaystyle \sum }_{\mathit{i}=\mathit{0}}^{\mathit{i}=\mathit{n}-\mathbf{1}}{\mathit{H}}_{\mathit{s}}\mathit{x}\mathit{C}$$

(2)

The blockchain ${\mathit{B}}_{\mathit{c}}$ size at the time ‘t’ can be evaluated approximately as shown in Equation (3):

$${\mathit{B}}_{\mathit{s}\mathit{i}\mathit{z}\mathit{e}}={\mathit{B}}_{\mathit{t}\mathit{i}\mathit{m}\mathit{e}}+{\displaystyle \sum }_{\mathit{i}=\mathbf{0}}^{\mathit{i}=\mathit{n}-\mathbf{1}}{\mathit{B}}_{\mathit{t}\mathit{x}\mathit{n}}^{\mathit{i}}$$

(3)

The evaluation of communication overhead in the proposed system is as follows

$${\mathit{C}}_{\mathit{o}}={\displaystyle \sum }_{\mathit{i}=\mathbf{0}}^{\mathbf{8}}\mathit{I}\mathit{n}\mathit{s}{\mathit{t}}_{\mathit{i}}\ast \mathit{B}{\mathit{C}}_{\mathit{t}}$$

(4)

where ${\mathit{C}}_{\mathit{o}}$ is communication overhead, which depends on generating the frequency of instances and processing them over a blockchain. In the process, ‘10′ random instances are considered from the MEHEALTH dataset and the communication overhead is evaluated using the proposed simulator. Generally, the total time to have the transactions in a block depends on the block’s gas limit and system workload. It also depends on the consensus mechanism that one has considered for authenticating and validating a transaction on a blockchain. The proposed application takes 1 KB to process the instances using the layers approach as discussed in Section 3. Furthermore, 0.030 ms has been consumed for processing an instance.

Table 3. Comparison of overhead with the existing systems.

No. of Instances	Communication Overhead (KB)
	IoMT _PLM [2]	GFB_CNN [34]	IoT_CMS [1]	Proposed System BGF_Blockchain
10	40	20	30	10
20	80	40	60	20
30	120	60	90	30
40	160	80	120	40
50	200	100	150	50
60	240	120	180	60
70	280	140	210	70

and Figure 4 show the comparison with the existing adaptive systems with the visualization.

Table 4. Parameters of GWO-PSO Hybrid model.

Learning Rate	0.1
Population size	Input Data from sensors
Hidden units	(10,15)
Batch size	(32,256)

represents different parameters to be considered while optimizing the Hybrid GWO-PSO model. Figure 5 represents the bar graph before optimizing the BGF-CNN model and after adding GWO-PSO optimization. After optimization, the prediction rate is increased gradually and the error rate is decreased. In addition to this, optimization with the Hybrid GWO-PSO model reduces the time complexity.

In comparison to the GFB-CNN technique [28], IoT-CMS [1] based system, and IoMT-PLM system [2], the suggested BGF-based system overhead is decreased by 50%, 60%, and more than 80%, respectively, as shown in Figure 5.

Table 5. Instances Gas costs.

No.of Instances	Gas Limit (Units)	Gas Cost	Gas Price CGWEI	Total ETH
10	205,430	1,276,220	1000	1.27622
20	402,860	2,652,440	2000	5.00488
30	518,290	4,018,660	3000	11.38598
40	805,720	5,114,880	4000	20.01952
50	1,132,150	6,071,100	5000	32.4055
100	2,054,300	12,762,200	10,000	126.622

demonstrates that the gas cost generates certificates with a 15-second block generation time without accounting for block header overhead. Figure 6 visualizes the results of Table 5.

Gas consumption costs for processing the instances in a block over an Ethereum blockchain are shown in Table 4 and visualized in Figure 6. The proposed system can be efficiently used to resolve the issues of preserving medical data securely. To do this, processed patient data is preserved securely over a blockchain with the inherited features of blockchain such as being more truthful, verifiable, transparent, and immutable.

Table 6. Comparison with the existing system by specified characteristics.

Approach	Instance Latency Time	Instance Process Time	Storage	System Type	Pre-Process	Application
IoT_CMS [1]	3.01	3.145	Cloud	IoT	CMS	Everyday Energy Saving
IoMT_PLM [2]	2.98	2.345	Cloud	IoT	PLN	Health care products
ECG_CNN [4]	2.78	4.34	Cloud	IR-UWB	CNN	ECG monitoring
GFB_CNN [34]	1.94	1.555	Cloud	CPS	CNN	Health Care
Proposed System	3.25	6.70	Blockchain	CPS	CNN	Patient Data

shows the proposed application’s performance by considering non-functional characteristics such as latency time and processing time. The results indicate the variance between the BCT platform and the non-BCT Platform. The time consumption is more on the BCT platform than on the non-BCT platform due to its internal computations such as mining, crypto hash evaluation, transaction and block creation, and joining of concern blocks into a blockchain.

Figure 7 shows the performance analysis of the proposed system with the existing approaches by considering the processing time and latency time, which are mentioned in Table 6.

Figure 8 shows the performance analysis of the proposed DL model with existing DL models by comparing time-lapse and processing times.

Detailed information on the transactions performed using blockchain is shown in Figure 9. This data consists of information relevant to heart signals, blood pressure, unhealthy heart signals, etc. Information can be retrieved from various parameters like block number, transaction (txn) hash, from address, to address, and the value of Ether, TxN fee, and Nonce, as shown in Figure 9.

5. Conclusions

Secure preservation is a major concern in all applications. Bayesian Grey Filter and convolution neural network are used in the proposed robust smart IoT sensors-based cyber-physical system. There are various security loopholes in the existing systems when compared to the proposed system. The proposed system gives a good accuracy which shows a positive notion of the performance factor. Furthermore, the results are preserved using blockchain technology, which has inherited features like immutable, transparent, crypto hash-based connectivity, etc. The proposed system is also compared with the existing system by considering some of the parameters like communication time, overhead time, latency time, and instance process time. The proposed system reduces the communication and overhead time with the help of the Bayesian Gray filter, CNN, and enhances the security using blockchain technology. In the future, the proposed system can be extended using HyperLedger blockchain technology to enhance security. The proposed work can also be extended into the fields like insurance and finance management [33]. It can be used in FinTech applications [34], which focus on using innovative technology solutions for improving financial services. Blockchain can manage and store a vast range of personal data securely.