Blockchain-Enabled Secure Data Sharing with Honey Encryption and DSNN-Based Key Generation

Abstract

Ensuring data confidentiality is a critical requirement for modern security systems globally. Despite the implementation of various access-control policies to enhance system security, significant threats persist due to insecure and inadequate access management. To address this, Multi-Party Authorization (MPA) systems employ multiple authorities for authorization and authentication, utilizing blockchain technology to store and access data securely, ensuring immutable and trusted audit trails. In this work, we propose a hybrid key-generation approach called the Identity and Attribute-Based Honey Encryption (IABHE) Algorithm combined with Deep Spiking Neural Network (DSNN) denoted by IABHE+DSNN for secure data sharing in a multi-party blockchain-based system. This approach incorporates various entities and multiple security functionalities to ensure data security. The data-sharing process involves several steps: initialization, authentication, initial registration, data protection, validation, and data sharing. Data protection is executed within the MapReduce framework, with data encryption performed using IABHE and key generation managed by DSNN. Experimental results demonstrate that the proposed IABHE+DSNN approach achieves a decryption time of 10.786 s, an encryption time of 15.765 s, and a key complexity of 0.887, outperforming existing methods.

1. Introduction

The increasing growth of Internet technologies has increased the flexible lifestyle of humans in day-to-day life and networked more social activities. In collaborative learning, some people might act dishonestly by giving false information or doing things wrong, either to keep data private or for selfish reasons [1,2]. There are many sensitive pieces of data created by social activities, like network activities, travel data, electronic health records, financial data, and personal information [3,4].

As cyber-physical systems become more linked to the Internet, they receive information from various sources at increasingly higher levels. This rising connectivity, complexity, and accessibility make these systems more prone to attacks [5,6]. In recent years, big data and different digital assets have received huge attention in this modern society due to technological advancements [7,8]. With the development of big data, several networks are being used to effectively store various kinds of sensitive data that are generated by different private and public sectors. The increasing demand for sharing sensitive data among enterprises and institutes has triggered social and economic benefits [9]. Big data creates large quantities of data and analyzes huge amounts of data in different engineering and scientific domains [7]. The data should be collected and validated regularly in the big data environment to enhance the security of sensitive information [10].

Generally, organizations and companies utilize the collected data to predict future trends, increase decision-making optimization, personalize services, and so on [8]. The security of the stored big data is a major factor in industry and various organizations to prevent unauthorized access [9,11]. Despite many applications and advantages, big data possesses many changes that should be effectively tackled to increase the quality of service. For instance, privacy and security issues, analytic management, and so on [12,13]. Over the past decades, due to the advancement of digitalization, highly sensitive data has been stored on cloud computing platforms rather than on paper. Various institutions, healthcare systems, and private and public organizations utilize cloud computing platforms for information storage [11,14]. Hence, various research studies have been performed by scholars to safely store sensitive information and provide effective access to data stored on cloud platforms. Presently, various services are provided by cloud computing technologies to effectively store and analyze big data. Cloud computing technologies effectively provide service support to analyze and store big data, which also enables real-time access and sharing due to the storage of more and more data in the cloud. Meanwhile, various issues, like the leakage of cloud data and data security issues, have arisen [3]. The security of the data is enhanced by storing the big data in a cloud or distributed file system [12]. However, the distributed data storage system dramatically increases the burden of protection on storage nodes due to difficulties in key management [15,16]. Thus, it is important to ensure data confidentiality, which is considered a major requirement of today’s systems. At present, the security and confidentiality of sensitive data are ensured by employing access-control policies. However, insecure and improper privileged access management poses various security threats, which also act as points of compromise in the system. Hence, MPA services are deployed to address these issues. Multiple authorities are utilized by MPA to perform authorization and authentication [17,18]. In general, MPA technologies effectively secure sensitive data from insider attacks by enabling the system to apply two different keys [17]. Here, the data are shared in a distributed environment, and the identity is managed to create a stable trust system for multiple parties to protect the user’s privacy and resist malicious attacks [19]. Furthermore, encryption is used to increase the security and privacy of shared and stored sensitive data to a certain extent by the data center. During encryption, the data owner encrypts the data, and the encrypted data are decrypted by authorized users [20,21]. Moreover, a new session key is negotiated in advance with a group of users to securely share and access the stored data. However, a new session key is required for encryption and negotiation of data if a new user is added to the authorized sharing group [22,23]. In recent years, data-sharing solutions using blockchain have been deployed to enhance security [24,25,26]. Here, the encrypted data are securely stored in the off-chain data center and records meta and data transfer logs in the blockchain with data auditing and retrieval [27]. Blockchain acts as a decentralized trust guarantee scheme, which utilizes a distributed ledger to record the operational results of each participant without tampering during cooperative training [28].

Blockchain technologies are utilized by various organizations and industries to enhance security and performance [29,30,31]. Blockchain effectively prevented the malicious participants from executing dishonest operations by specifying the endorsing node of each participant in the cooperative training [32,33,34]. Generally, blockchain has a revolutionary impact on the secure sharing of sensitive information, which also provides a distributed platform with no centralized authority support in different industries [32,35]. Deep learning techniques have been utilized in recent years to provide full security authentication by securely sharing the data over the network [36,37].

This article presents the Identity and Attribute-Based Honey Encryption (IABHE) combined with the Deep Spiking Neural Network (DSNN) model (IABHE+DSNN) for secure data sharing in a blockchain-based multi-party system. The IABHE algorithm ensures high security and confidentiality by encrypting data based on user identity and attributes, while Honey Encryption (HE) provides plausible decoys for incorrect decryption attempts, enhancing security against brute-force attacks. The DSNN component improves the key-generation process, making it more secure and reliable. The proposed model employs a permissioned blockchain framework to ensure secure, scalable, and efficient data sharing. Incorporating blockchain technology into IABHE+DSNN for secure data sharing in a multi-party system enhances the overall contributions of the work. By leveraging blockchain technology, the system benefits from its inherent characteristics, such as decentralization, transparency, and security. It involves entities like data requesters, data owners, InterPlanetary File System (IPFS), encryption servers, and Multi-Party Authorization (MPA) systems. Various security functionalities, including hashing, encryption, session passwords, and One-Time Passwords (OTP), are integrated. The data-sharing process comprises initialization, data protection, authentication, registration, validation, and sharing phases. During initialization, security functionalities are set up, and entities are registered. Authentication enhances user security, and data are encrypted using IABHE in the MapReduce framework, with DSNN generating the secret keys. The reducer phase aggregates encrypted data through polynomial interpolation. As big data continues to grow, securely sharing and storing sensitive information is critical. Current schemes often lack trust and rely on centralized access. This motivates the development of our innovative approach to secure data sharing.

The key contributions of this work are given as:

• We design an algorithm called IABHE+DSNN for securing a multi-party data-sharing system. It was developed to securely share the data via MapReduce framework a multi-party data-sharing system.
• In this work, data encryption is executed using the IABHE model, and a secret key is developed for data protection using the DSNN model.
• We use blockchain technology for storing and accessing data. This approach integrates data protection and transparency of data transactions.

The remaining sections of the article are arranged as follows: Section 2 depicts the baseline techniques used for secure data sharing in multi-party data-sharing systems, and Section 3.1 portrays the system model. Moreover, the developed IABHE+DSNN for secure data sharing is delineated in Section 3.2. Furthermore, Section 4 demonstrates the outcomes as well as the discussions. Finally, Section 5 portrays the conclusion of the article.

2. Related Work

In recent years, researchers have focused on encryption techniques using blockchain technology. For example, Wang et al. [3] designed blockchain + Proxy Re-Encryption (PRE) + Trusted Execution Environments (TEEs) to increase the security of data-driven systems for protecting sensitive data. This approach effectively reduced the communication and computation overhead issues that occurred while securely sharing data in data-driven systems. However, it was not successful in providing security to the data by increasing data confidentiality and maintaining privacy with data flow control in real time. Singh et al. [7] developed a Medi-Block record for the protection of medical records shared with patients and hospitals. This model significantly eliminated third-party trust issues and utilized the concept of bilinear mapping for secure data sharing in the authentication phase. It significantly reduced the average communication time by meeting the security requirements of data record sharing. Meanwhile, the total communication cost of the model increased linearly while medical data was being retrieved by providing full security authentication. Yin et al. [9] devised a function encryption-based privacy-preserving method for providing solutions to trust issues among various participants. This model was highly robust in multi-party data sharing for privacy protection but was not suitable for single-party authorization for secure sharing of data.

Chen et al. [10] established a blockchain-based multiauthority revocable CP-attribute-based encryption (MA-RABE) scheme for secure data sharing. It converged quickly while sharing data securely to attain highly efficient user revocation and to ensure the privacy policy of cloud servers. However, it encountered difficulties in reading and deleting the data in the system while securely sharing the data via a multi-party data-sharing system. Bakir [12] introduced a blockchain-based Special Key Security Model (BSKM) to enhance data confidentiality by controlling the data flow and increasing privacy. This technique effectively performs blockchain transaction operations to ensure data consistency, integrity, and confidentiality with less computational time. Meanwhile, this approach failed to consider multi-function keys by considering user groups to increase the security and privacy of big data-based financial and bank data. Alhazmi et al. [14] developed a fragmentation method for increasing the security of big data systems. This technique avoided the overhead of encryption for non-sensitive as well as low-data portions to increase big data security. However, it failed to implement the solution using the Hyperledger fabric platform to promote high transaction security and throughput. Qin et al. [15] designed blockchain and trusted data cloud centers and PairHand to increase the security of cloud data centers. This approach reduced the burden on cloud centers during data processing and increased the difficulty for illegal intruders in obtaining data by storing block mapping information to ensure the security of data storage. However, it failed to consider data visual validation techniques to identify the presentation and mining of data value and to determine the patterns among the data attributes for enterprise decision-making.

Battah et al. [17] established a fully decentralized blockchain-based solution to provide access to shared encrypted data stored on decentralized and public storage platforms. This approach attained high generalizability by making the components of the system optional during the exchange of the data policy of prioritized and confidential data securely in the larger system. Satyabrat et al. [38] introduced methods to efficiently outsource modular exponentiation computations to the cloud, aiming to alleviate the computational burden on resource-constrained IoT devices. They present innovative solutions to enhance the performance and security of IoT applications. Scalability challenges may arise when deploying the approach in large-scale IoT environments, requiring further analysis and optimization. However, this approach failed to fully mitigate some security concerns in public and decentralized storage platforms, such as Denial of Service (DoS) attacks, by limiting the maximum number of responding oracles. To enhance fairness and transparency in blockchain protocols, Caldarola et al. [39] proposed work on the Neural Fairness Blockchain Protocol Using an Elliptic Curves Lottery. Their work, which promises the advancement of blockchain technology but has drawbacks, is essential for assessing its viability and ensuring its alignment with the goals of fairness, transparency, and decentralization in blockchain systems.

Challenges

The limitations of various baseline models utilized for secure data sharing are given below.

• The PRE + TEE technique used in [3] significantly reduced the reduplicative data as well as the ciphertexts, but it failed to consider computational complexity problems that occur while sharing data securely in data-driven systems. The complexity problems are evaluated by generating the key using DSNN.
• The Medi-Block record scheme utilized in [7] attained less storage overhead while providing a secure platform for medical data sharing. However, third-party storage services are required to share medical data securely. The hashing parameters and interpolation used in the proposed research ensure the security of data sharing.
• The MA-RABE method employed in [10] was highly effective and robust in providing security for multi-party data sharing, but it failed to decrypt local models even after obtaining the encryption and decryption keys of participants. The encryption and decryption algorithms enhanced in the proposed research ensure security in multi-party data sharing.
• The BSKM model used in [12] effectively stored the data for differential transactions and provided reliable data communication. Meanwhile, it recorded a high time for writing and updating transactions in objects while enhancing the privacy and confidentiality of data. The evaluation metrics that evaluated the encryption and decryption times show the confidentiality of the data.

Existing systems for secure data sharing may have limitations such as scalability issues with large datasets, lack of robust encryption methods leading to potential security vulnerabilities, and inefficiencies in key-generation processes. These drawbacks can hinder the effectiveness of data-sharing processes and compromise the confidentiality and integrity of shared information. The prevailing data encryption approaches effectively protect the privacy and security of stored, sensitive information shared through data centers. Meanwhile, these approaches suffered from increasing data auditability and transparency issues. They also required the deduplication of plaintext and secure management of decryption keys to secure data in data-driven systems.

Table 1. Methods, advantages, and limitations/disadvantages of existing work.

Authors & Reference	Methodology	Advantages	Limitations/Disadvantages
Wang et al. [3]	Security enhancements: Blockchain, PRE, TEEs	Robust security measures, safeguarding sensitive data	Lack of data visual validation techniques; High computational and storage overhead
Singh et al. [7]	Medi-Block record: Secure data sharing using blockchain technology. Utilizing blockchain-based authentication, bilinear mapping	Tamper-proof and anonymous identity management	Storage burden due to redundant data copies; Scalability issues in large networks
Yin et al. [9]	A blockchain-based collaborative training method for multi-party data sharing. Employing blockchain-based collaborative training, distributed ledger	Prototype system to analyze and evaluate time consumption	Limited success in enhancing data security; High latency in data processing
Chen et al. [10]	Industrial chain data sharing: Blockchain, Big Data Technology	Enhanced data sharing and circulation efficiency	Integration complexity, need for standardization; Difficult implementation
Bakir et al. [12]	New Blockchain-Based Special Keys Security Model With Path Compression Algorithm for Big Data. Utilizing a special key security model, path compression algorithm	Ensures data confidentiality, integrity, and consistency	Need for efficient public auditing techniques; High implementation cost
Alhazmi et al. [14]	Towards big data security framework by leveraging fragmentation and blockchain technology. Employing fragmentation techniques, lightweight metadata structure	Eliminates the need for third-party auditing	Complexity of implementation; Fragmentation may impact performance
Qin et al. [15]	Big data security architecture: Blockchain, Trusted Data Cloud Center	Enhanced security, integrity, and transparency of data	Potential scalability issues, complexity; High initial setup and maintenance costs
Battah et al. [17]	MPA: Blockchain-based authorization, IPFS Encryption	Secure multi-party data access, decentralized storage	Latency in data retrieval; Limited network performance
Rath et al. [38]	Secure Outsourcing: Parallel processing, secure cloud outsourcing	Efficient computation for IoT, reduced processing load	Security risks in outsourcing; Dependence on cloud services
Caldarola et al. [39]	Neural fairness protocol: Neural fairness blockchain protocol, elliptic curves lottery	Ensures fairness in transactions, enhanced security	Computational overhead; Complex implementation
Proposed System IABHE+DSNN	Data protection within the MapReduce framework, data encryption using IABHE, key generation managed by DSNN.	Enhanced data security through hybrid key-generation approach	Research will focus on addressing additional security concerns and analyzing systems within large network behavior

presents details such as technique, limitation, advantage, etc., of a few existing works.

3. The Proposed Methodology

3.1. System Model for Blockchain-Based Multi-Party Data Sharing

The system model for a blockchain-based multi-party data-sharing system comprises different entities that help to provide access to various encrypted data stored in IPFS [17]. Blockchain ensures decentralization by distributing the encrypted data across a network of nodes, eliminating the need for a central authority to manage transactions or access control. This decentralized architecture enhances resilience and mitigates the risk of single points of failure, making the system more robust and trustworthy. For blockchain-based multi-party data-sharing systems, the data requester, data owner, IPFS, and re-encryption oracles are used to perform communication to govern access control to encrypted data. A unique Ethereum address is provided to each entity for blockchain communication. The different entities utilized by the system are briefly delineated below.

3.1.1. Data Owner

The data owner serves as the initial point of the system and is responsible for tasks such as registering data addresses, understanding user requirements, facilitating communication, and enabling data sharing within the blockchain framework. Utilizing a symmetric key algorithm, the data owner encrypts the data before forwarding it to a decentralized peer-to-peer (P2P) database. Additionally, the data owner shares encrypted public keys with other authorized parties, including MPA (Multi-Party Agreement) participants, using multi-signature techniques. Furthermore, the data owner establishes a smart contract containing data addresses and hashed components to ensure transparency and accountability within the system. As part of the data-sharing process, the data owner generates a re-encryption key derived from the data requester’s public key and transmits this key to proxy servers for further processing and access control.

3.1.2. Data Requester

The Ethereum address is used by the data requester to access the encrypted data given by the data owner by contacting the smart contract. Then, the data requester waits after validating the requester to obtain token access from the smart contract to receive the data. Later, the hashed file, encrypted symmetric key, and data are downloaded by the requester after downloading data from the proxy. Finally, the data requester decrypts the data as well as the symmetric key by utilizing the private key and the symmetric key to decrypt the data again.

3.1.3. IPFS

The IPFS is considered a P2P decentralized database that helps to share data with multiple users. Here, the data owner uploads the encrypted data with a symmetric key, which is encrypted by the data owner’s public key. The IPFS provides encrypted data to the proxy with an encrypted symmetric key after the data are requested from the database.

3.1.4. Encryption Server

Encryption is considered a compute-intensive task that utilizes an Ethereum-based smart contract approach, which is very expensive. The proxy re-encryption servers fetch data and effectively perform complex functions. The encryption server helps to share data between the data owner and requester. Generally, a reputation system is presented on proxy servers that are managed by smart contracts. Moreover, a unique address is presented on the proxy server that is forwarded with access tokens shared by the requester. Likewise, a token is received by the proxy server with the address of the requester to execute the validation task. Data privacy, integrity, and confidentiality are ensured after the requester performs the data-sharing task. The proxy server initially receives the re-encryption key from the data owner and downloads the data, which involves encrypted data and a symmetric key from the decentralized dataset. Once the proxy server comprises data and a key, the re-encryption of the symmetric key is performed, and the key is forwarded to the requester.

3.1.5. MPA

In general, the MPA acts as co-owner and is presented in each phase of the access-control mechanism. The MPA effectively avoids malicious acts by managing access to shared wallets. The keys of MPA are used by the data owner through the utilization of multi-signature technologies. Moreover, the MPA entities are qualified to check the essential requirements for providing access to data requested by the data receiver. The MPA also helps to secure highly sensitive data against insider attacks. The system model of the multi-party data-sharing system is displayed in Figure 1.

3.2. The Proposed Map-Based IABHE+DSNN

This article presents an IABHE+DSNN model for secure data sharing using the MapReduce framework. At first, the system model of blockchain-based multi-party data sharing is contemplated, along with several entities, namely IPFS, data owner, data requester, encryption server, and MPA. These are entities creating a chain of blocks that are resistant to manipulation or unauthorized access. The IABHE+DSNN model is designed by deriving a mathematical model that utilizes different security functionalities such as hashing, encryption, session passwords, OTP, etc. The steps followed for securing big data are system initialization, registration, authentication, data protection, validation, and data sharing. The initial phase of the model involves system initialization, where various security parameters are set up for subsequent operations. Following this, registration of the data owner, data requester, MPA, and encryption server occurs. Subsequently, authentication is conducted to confirm the user’s identity. In the data-protection phase, the data encryption server safeguards stored data. The protected data are then stored in the IPFS, and validation is performed to verify the requester’s identity. Additionally, during the data-protection phase, the input data undergoes processing in the MapReduce framework, comprising the mapper and reducer phases, to construct the data encryption model. Encryption using IABHE with a key-generation process is carried out in the mapper phase. The IABHE is designed by modifying the encryption algorithm given in [40] by incorporating identity-based encryption, and the secret key is generated using the DSNN [41] model. Furthermore, the efficiency of accumulated encrypted data is enhanced in the reducer phase by executing polynomial interpolation. Finally, the data are shared with the user once the validation is completed. The systematic view of the IABHE+DSNN model for secure data sharing in the MapReduce framework is portrayed in Figure 2. The detailed interconnections and relationships are given below:

• Blockchain Network: In this system, the blockchain network acts as the central controller and orchestrates the interactions among the various components, including the subcomponents in both the blockchain-based multi-party network and the MapReduce framework. It is responsible for managing the overall workflow and system phases, such as initialization, registration, authentication, data sharing, validation, and data protection.
• IPFS: Connects with IMPEncryption Server to store encrypted data and interacts with the data requester to provide stored encrypted data.
• DataOwner: Uses the MPEncryptionServer to encrypt data. Stores encrypted data in IPFS and shares data through a blockchain network.
• DataRequester: Requests data through BlockchainNetwork. Retrieves encrypted data from IPFS and uses IMPEncryptionServer to decrypt data.
• IMPEncryption Server: Performs encryption and decryption using keys generated by DSNN and works with MapReduce Framework to handle large-scale data encryption tasks.
• MapReduce Framework: Utilizes IABHE for encryption operations and depends on DSNN for key generation.
• IABHE: It encrypts data within the MapReduce framework and generates the necessary encryption keys.
• DSNN: Generates secret keys for encryption and decryption.
• System Phases: Encapsulates the phases managed by the blockchain network and Coordinates activities across the entire system.

3.2.1. System Initialization

The initialization of random numbers, security parameters, and public keys is executed in MPA during initialization. Here, u and v are the random numbers fixed to 0 and 2, the security parameters utilized for secure data sharing are indicated as A and B, and $X_K$ resembles a public key. Moreover, Figure 3 shows the initialization of parameters in MPA for secure data sharing.

3.2.2. Registration

The registration is carried out among the data owner and MPA, as well as between the data requester and MPA. The entities, namely the data requester, data owner, MPA, encryption server, and IPFS, are used to perform the registration task in the registration phase. The registration performed during registration is elaborated as follows: Initially, the registration is carried out among the data owner and MPA for secure data sharing. Here, the $ID$ $DO\_ID$ and password $D{R}_{P}WD$ of the data owner are fed into MPA and are stored as ${\mathrm{DO}\_\mathrm{ID}}^{*}$ and ${\mathrm{DO}\_\mathrm{PWD}}^{*}$. Later, a message $E_1$ is created by hashing the stored data owner password ${\mathrm{DO}\_\mathrm{PWD}}^{*}$ concatenated with security parameter A, and this hashed value is concatenated with random number u concatenated along with the public key of data owner $X_K^O$. The generated message $E_1$ is expressed as,

$$E_1=h({\mathrm{DO}\_\mathrm{PWD}}^{*}\oplus A)\left|\right|u\left|\right|X_K^O$$

(1)

(a)

Registration between the data owner and MPA: The generated message is sent to the data owner, and this generated message $E_1$ is stored in the data owner as ∼$E_1$. Finally, this stored message in the data owner ∼$E_1$ is again forwarded to MPA and stored for registration. The data owner is registered in MPA if the created message $E_1$ is equal to the stored message in data owner ∼$E_1$. Figure 4 shows the registration of the system among the data owner and MPA.

(b)

Registration between data requester and MPA: The registration among the data requester and MPA is performed by initially passing the data requester ID $D{R}_{ID}$ and password $D{R}_{PWD}$ to MPA, where it is stored as $D{R}_{ID}^{*}$ and $D{R}_{P}W{D}^{*}$. The MPA forwards a copy of these data to the IPFS, and the details are stored as $D{R}_I{D}^{*}$ and $D{R}_{P}W{D}^{*}$ in IPFS. Then, a message $E_2$i is generated in MPA by XOR-ing the hashed value of stored data requester password $D{R}_{P}W{D}^{*}$ concatenated with security parameter B along with the public key of the data owner $X_K^R$, which is given by,

$$E_2=h(D{R}_{PWD}^{*}\Vert B\mid )\oplus X_K^R>$$

(2)

Later, the created message $E_2$ is fed to the data requester and stored $E_2$ in the data requester, and this stored message is forwarded to MPA for registration. If the created message $E_2$ is equal to the stored message $E_2$ in the data requester, then the data requester is registered with the MPA. Figure 5 depicts the process executed during registration among the data requester and MPA.

3.2.3. Authentication

Generally, authentication is performed to promote quick and easy accessibility to the resources. In this authentication phase, the authentication is carried out between the data owner and MPA as well as between the data requester and MPA. The authentication process carried out is elaborated below,

(a)

Authentication between the data owner and MPA: The authentication request message $F_1$ is generated by the data owner during authentication between the data owner and MPA. The message $F_1$ is created by XOR-ing the hashed value of random number concatenated with the recorded data owner public key ${⟦X_K^O⟧}^{*}$ concatenated with the modulus of the security parameter A along with the hashed value of data owner password $D{O}_{P}WD$. Thus, the authentication request message $F_1$ generated is given as,

$$F_1=h(u\Vert {X_K^O}^{*}\mid \Vert \mathrm{mod}A\mid )\oplus h\left(D{O}_{PWD}\right)$$

(3)

The generated authentication request in the data owner is forwarded to MPA, and MPA generates the message $F_1$ using the credentials available with it. The message $F_1$ is obtained by XOR-ing the hashed value of random number u and concatenating it with data owner public key $⟦X_K^O{⟧}^{*}$ concatenated with the modulus of the security parameter A and hashing the value of stored data owner password $D{R}_{P}W{D}^{*}$. Thus, the generated message $F_1$ is designated as,

$${\tilde{F}}_1=h(u\Vert X_K^O\mid \Vert \mathrm{mod}A\mid )\oplus h\left(D{O}_{PWD}^{*}\right)$$

(4)

The data owner is verified in MPA if the authentication request message $F_1$ in the data owner is the same as the recorded message $F_1$ in MPA. Then, an OTT is generated in MPA to authenticate with the data owner. The OTT is created by XOR-ing the hashed value of recorded data owner ID $D{R}_I{D}^{*}$ and data owner public key $⟦X_K^O{⟧}^{*}$, which is expressed as

$$OTT=h(D{O}_{ID}^{*}\Vert A\mid )\oplus X_K^O$$

(5)

Then, the generated OTT in MPA is passed to the data owner and stored as $OTT$, where $OTT$ is generated by XOR-ing the hashed value of data owner ID $D{O}_{I}D$ and security parameter along with the data owner public key $⟦X_K^O{⟧}^{*}$ and is given by,

$$O\tilde{T}T=h(D{O}_{ID}\Vert A\mid )\oplus {X_K^O}^{*}$$

(6)

The $OTT$ generated by the data owner is forwarded to MPA for authentication. If the created OTT in MPA is equal to the generated $OTT$ in the data owner, then the data owner is authenticated in MPA. Moreover, the process performed during authentication among the data owner and MPA is displayed in Figure 6.

(b)

Authentication between the data requester and MPA: An authentication request message is generated by the data requester for authentication between the data requester and MPA. Here, it is created by XOR-ing the hashed value of the data requester ID and concatenated with a random number along with the hashed value of the data requester password concatenated with a security parameter. The generated authentication message is expressed as,

$$F_2=h(D{R}_{ID}\Vert u\mid )\oplus h(D{R}_{PWD}\Vert B\mid )$$

(7)

Then, the message created by the data requester is sent to MPA for verification. On receiving the authentication message, the message is created by the MPA by XOR-ing the hashed value of the data requester ID, concatenating and concatenated with a random number hashing the value of the data requester password, and then concatenating with a security parameter. The created message is expressed as

$${\tilde{F}}_2=h(D{R}_{ID}^{*}\Vert u\mid )\oplus h(D{R}_{PWD}^{*}\Vert B\mid )$$

(8)

Moreover, the data requester is verified to see if the authentication message created by the data requester is equal to the generated message in MPA. Later, an is generated in MPA by hashing the XOR-ed value of the data requester password and random number, which is expressed as

$$OTP=h(D{R}_{PWD}^{*}\oplus u)$$

(9)

The data generated in MPA is sent to the data requester and recorded, and the recorded data requester is again passed to MPA for verification. Here, the data requester is authenticated with MPA if the stored data are the same as the recorded data. Figure 7 displays the authentication performed among the data requester and MPA.

Data protection is generally performed on the encryption server in a MapReduce framework. Here, data encryption is accomplished using IABHE, and a secret key is generated using DSNN [40] in the mapper phase. Moreover, the resultant encrypted data are aggregated using polynomial interpolation in the reducer phase, and the encrypted data are published in the cloud. Furthermore, the resultant output obtained from the map function will be key-value pairs. The coefficients of the polynomial are taken as values, and the key is the interpolated degree of the polynomial. The process carried out during the encryption of data and key generation is briefly demonstrated as follows.

3.2.4. Key Generation Using DSNN

The key generation is accomplished by applying the data to the DSNN, and the data in the source domain is transferred to the transformation domain. Furthermore, the data are considered to be the seed for generating the secret key. Here, the style of the secret key depends on the transformation domain. The key is generated by taking into account the inner layers of the DSNN during the training process. The DSNN [41] model is an event-driven and data-driven hierarchical network that transmits signals as spikes among neurons. DSNN plays a crucial role in secret key development by providing a robust and efficient method for generating encryption keys. The use of neural networks enhances the security of the encryption process and ensures that the generated keys are complex and difficult to decipher. The DSNN comprises a similar structure to Convolutional Neural Networks (CNN) that replaces the nonlinearities of spiking neurons to effectively perform the key-generation task. The DSNN combines both the advantages of SNN and CNN and determines the network parameters by training CNN and converting it to SNN. Here, the CNN is trained initially to obtain network parameters, like bias and weight, and then converted to the corresponding layer of SNN. It generally comprises integrate-and-fire neurons and performs convolution, max pooling, normalization of weights, and realization of biases. Moreover, CNN and SNN are adjusted to decrease the conversion losses that occur during conversion. The process carried out in each layer is demonstrated below,

• CNN: CNN is a multi-layer supervised learning neural network that mainly comprises two core models, such as convolution and pooling layers, for feature extraction from the input data. The convolution and pooling layers are adjusted to reduce the dimension of the data and to extract abstract features from the data. Thus, the resultant data from CNN is fed into SNN for the generation of a secret key for the encryption task.
• Integrate-and-fire neuron model: The DSNN utilizes a simple integrate-and-fire neuron model for key-generation task.
  The integrate-and-fire neuron dynamics are expressed by,

  $$\frac{d{v}_{\mathrm{mem}}\left(Q\right)}{dQ}=\sum _x\sum _{h\in C_x}{I}_x\beta (Q-h)$$

  (10)
  Here, the Dirac function is represented as $\beta (.)$, the synapse weights of $x^{\mathrm{th}}$ input neuron are indicated as $C_x$, the spike time series is denoted as $I_x$, and Q resembles time.
• Convolution operation: The convolution operation carried out in DSNN is the same as of CNN, which is designated as,

  $$U=\sum _x{I}_x{a}_x$$

  (11)

  where the input given to the convolution kernel is given by $a_x$, and the output is symbolized as U.
• Max pooling operation: The max pooling operation in DSNN cannot be performed using a simple maximum value operation due to the transmission of discrete spike signals by DSNN. Thus, global average pooling is performed in DSNN to perform the max pooling operation.
• Weight normalization: In the DSNN model, the fire rate of spiking neurons is ensured by multiplying the spike neuron weights using weight normalization. The weight normalization coefficient is selected using a robust normalization method. Here, the spiking activation is performed to measure the weight normalization coefficient.
• Realization of biases: The data range is updated by adding biases to the network, where the constantly released spike signals by the neurons are utilized as biases. In general, the weights of connections among neurons are considered to change the bias values, and this process is repeated to stimulate biases.

Thus, the secret key K is finally created from the data W by DSNN during the key-generation phase. Here, the secret key corresponds to an intermediate layer of the DSNN. During the training process, the parameters of the internal layer change, and the key generated is considered to be one-time paid. Figure 8 shows the structure of DSNN.

3.2.5. Data Encryption Using IABHE

IABHE enhances security by combining identity and attribute-based encryption to provide fine-grained access control [36]. Honey encryption creates dummy data to trick uninformed decryption efforts. Attribute-Based Encryption (ABE) was created to give flexible access control based on user attributes, while honey encryption was invented to create convincing-looking decoys to protect against brute-force results. In this proposed system, first, the IABHE is applied in the mapper phase for data encryption by identifying the seed space, as shown in Algorithm 1 [36]. The algorithm for secure data sharing using the IABHE combined with DSNN involves several key steps to ensure data confidentiality and secure sharing. Here is an explanation of each step in the algorithm:

• Initialization: The process begins with initializing the system and setting up the necessary parameters for data encryption and key generation. This step prepares the algorithm for secure data-sharing operations.
• Authentication: Authentication involves verifying the identities of the parties involved in the data-sharing process. This step ensures that only authorized entities can access and share data securely.
• Initial Registration: During initial registration, entities are registered within the system and provided with the necessary credentials for data access and sharing. This step establishes the foundation for secure data sharing among multiple parties.
• Data Protection: Data protection is a crucial step where sensitive information is encrypted using the IABHE. This encryption process ensures that data remain confidential and secure during storage and transmission.
• Validation: Validation involves verifying the integrity and authenticity of the encrypted data to prevent unauthorized access or tampering. This step ensures that only authorized parties can decrypt and access the shared data.
• Data Sharing: The final step in the algorithm is data sharing, where encrypted data are securely transmitted between authorized parties using the generated keys from the DSNN. This step facilitates the secure and efficient sharing of sensitive information among multiple entities.

Algorithm 1 Mapper-Based IABHE Algorithm

1: Input: Data Owner (DO)/Data Requester (DR) registration data   2: System Initialization:   3: Initialize random number generator   4: Define security parameters   5: Generate public key   6: Registration Phase:   7: Encrypt DO registration data: $E_1=h(DO\_PW{D}^{*}\oplus A)\parallel u\parallel X_K^O$   8: Encrypt DR registration data: $E_2=h(DR\_PW{D}^{*}\parallel B)\oplus X_K^R$   9: Authentication Phase: 10: Calculate $F_1$ for DO authentication 11: Verify DO authentication: $OTP=h(DO\_I{D}^{*}\parallel A)\oplus X_K^O$ 12: Calculate $F_2$ for DR authentication 13: Verify DR authentication: $OTP=h(DR\_PW{D}^{*}\oplus u)$ 14: Key-Generation Phase: 15: if Using DSNN then 16:     Calculate gradients for key generation: $\frac{dv\_mem\left(Q\right)}{dQ}={\sum }_x{\sum }_{h\in C_x}{I}_x\beta (Q-h)$ 17: else 18:     Use alternative key-generation method 19: end if 20: Data Encryption Phase: 21: if Using IABHE then 22:     Data Owner sends data to the encryption server 23:     Encrypt data: $J=HE(W,k)$ 24:     $G\leftarrow \mathrm{encode}\left(W\right)$ 25:     $M\leftarrow {\{0,1\}}^n$ 26:     $G^{\prime }$$\leftarrow H(M,k)$ 27:     $N\leftarrow G^{\prime }\oplus G$ 28: else 29:     Use alternative encryption process 30: end if 31: Validation and Data-Sharing Phase: 32: if Data decryption is needed then 33:     Decrypt data: $D\left(K(M,J^{***})\right)$ 34:     $G^{\prime }\leftarrow H(M,K)$ 35:     $G\leftarrow N\oplus G^{\prime }$ 36:     Decode decrypted data: $W=\mathrm{Decode}\left(G\right)$ 37: else 38:     Skip decryption process 39: end if 40: return Encrypted data: $(M,N)$ / Decrypted data: W

The IABHE is used to encrypt the input data, where the data encryption is performed by utilizing the honey encryption approach [40], which follows different processes, like the Distribution Transforming Encoder (DTE), password distribution, and honey words for message distribution. The encryption process is established by considering the attributes as well as the identity of the user. The honey encryption approach is carried out in both speed space and message space. Speed space and message space refer to the efficiency and capacity of a cryptographic system in terms of processing speed and message size. Speed space (how quickly data can be encrypted and decrypted) and message space (the size of data that can be securely transmitted) are used to evaluate the performance and scalability. The honey encryption algorithm performs the encryption process under the honey encoding and decoding model. Moreover, the DTE is used in speed space to map all messages based on a step-by-step process, where DTE effectively processes both the encoding and decoding processes. The message is encrypted by identifying the possible message space initially during the honey encryption process, and the messages are arranged in a particular order. Then, the cumulative probability of all messages is measured in the message space. Later, the messages are effectively mapped using DTE in message space, and the ciphertext is finally generated by XOR-ing with a key.

In the data-protection phase, the cryptographic hash function is utilized to encrypt the data and the secret key. Moreover, a seed is generated by a uniform random assignment of the encoded data. A random string is generated by a uniform random assignment and is set to 0 and 1. Finally, the encrypted data are stored in IPFS, and Figure 9 displays the process performed in the data-protection phase, where the uniform random assignment is signified as $. IPFS is a decentralized storage network, which means that the data are not stored on a single server but are instead distributed across multiple nodes. This enhances data availability and resilience, as the data can be retrieved from multiple locations. The data distribution is handled by a uniform random assignment process, ensuring that the data are evenly spread across the network, which is crucial for maintaining the efficiency and reliability of the decentralized storage system. The uniform random assignment process ensures the data are evenly and randomly distributed, represented by $.

3.2.6. Validation and Data-Sharing Phase

The validation of the data requester in MPA is performed in the validation phase. Here, a data access request is sent by the data requester to MPA, and the validity of the request is checked by MPA. If validity is established, MAP forwards the secured data to the requester. The validation process performed between the data requester, MPA, and IPFS is briefly explained in this section. Initially, the data requester’s ID $D{R}_{ID}$ and password $D{R}_{PWD}$ are passed to MPA, which is stored as $D{R}_{ID}^{*}$ and password $D{R}_{PWD}^{*}$ in MPA. The data requester ID $D{R}_{ID}$ and password $D{R}_{PWD}$ are verified with the stored data requester ID $D{R}_{ID}^{*}$ and password $D{R}_{PSWD}^{*}$ in MPA. After successful verification, a session password $S_{PWD}$ is generated by the MPA by finding the hashed value of XOR-ed data requester ID $D{R}_{ID}$ and password $D{R}_{PWD}$ concatenated with the modulus of random number and XOR-ing it with the security parameter. The generated session password is expressed as,

$$S_{\mathrm{PWD}}=h(D{R}_{\mathrm{ID}}^{*}\oplus D{R}_{\mathrm{PWD}}^{*}\Vert \mathrm{mod}u\mid )\oplus A$$

(12)

The session password $S_{PWD}$ is forwarded to IPFS, which is stored as $S_{PWD}^{*}$, and the IPFS generates a time stamp T. Later, a message Y is created by the IPFS by hashing the recorded session password $S_{PWD}^{*}$ with time stamp T. The message Y generated in IPFS is given by,

$$Y=h(S_{\mathrm{PWD}}^{*}\Vert T\mid )$$

(13)

Afterward, the generated message Y in IPFS is passed to MPA and stored as $Y^{*}$, and the validation of the timestamp is performed. The MPA produces a message Y by hashing the generated session password $S_{PWD}$ concatenated with a time stamp T, which is given by,

$$\tilde{Y}=h(S_{\mathrm{PWD}}\Vert T\mid )$$

(14)

The created message Y is sent to the IPFS for verification, and the IPFS sends the protected data to MPA if the received message Y from the MPA is the same as the message Y generated in IPFS. Later, the data $J^{*}$ stored in IPFS is shared with MPA, and it is stored as $J^{**}$ in MPA. Subsequently, the stored data $J^{**}$ in MPA is shared with the data requester, where it is stored as $J^{**}$. The validation and data-sharing process performed among the data requester and MPA is given in Figure 10.

3.3. Data Decryption

The decryption is performed once the data are validated and shared with the data requester so that the data can be accessed. Here, data decryption is performed to easily access the relative data along with their attributes using an identity-based decryption key. The data requester decrypts the protected data by initially XOR-ing the ciphertext and key K. Later, the mapping of the seed G to the message is performed by inversely applying DTE to obtain plain text. Later, the decryption process is performed by applying aggregated encrypted data, and the data decryption process performed by the data requester is given in Figure 11. Here, $J^{***}$ represents the cipher text and M is a random string.

To clearly illustrate the entire process of our proposed IABHE+DSNN method for secure data sharing, we provide an example of the encryption and decryption process below.

• Sample Data:
  Original Plaintext: “Sensitive Data Example”
• Encryption Process:
  Encryption Key: A1B2C3D4E5F6G7H8
  Generated Ciphertext: f8h29jkwe823jhds8k
• Decryption Process:
  Decryption Key: A1B2C3D4E5F6G7H8
  Recovered Plaintext: “Sensitive Data Example”

The above-mentioned sample is derived from our test dataset, which demonstrates the transformation from plaintext to ciphertext and back to plaintext using our proposed method. This process highlights the security and reliability of the IABHE+DSNN approach, ensuring that data remains protected during storage and transmission.

4. Results and Discussion

The results obtained by the IABHE+DSNN model designed for securely sharing data in the MapReduce framework, as well as corresponding discussions performed to identify the performance, are elaborated as follows.

4.1. Experimental Setup

The IABHE+DSNN technique designed for secure data sharing is executed using PYTHON tool version 3.9.11 on a PC with OS-Windows 10, RAM-8 GB, ROM-More than 100 GB, GPU and CPU-1.7 GHz.

• Dataset description:
  The database considered for blockchain-based multi-party data sharing is taken from the Skin Segmentation database [42] and Localization Data for Person Activity database [43].

  (i)

  Skin Segmentation database: The Skin Segmentation database given in [42] comprises about 245,057 learning samples, where 50,859 samples are skin samples and 194,198 are non-skin samples. Moreover, the data are collected randomly from B, G, and R values from different genders, race groups, and age groups.

  (ii)

  Localization Data for Person Activity database: Data for Person Activity database: The Localization Data for Person Activity database [43] possesses the various activities of five people, and the data are recorded from four tags, such as chest, belt, ankle right, and ankle left. Moreover, the tags are determined by one of the attributes.

• Evaluation measures:
  The different parameters, like key complexity, decryption time, and encryption time, are utilized to identify the performance of the IABHE+DSNN approach used for secure data sharing, and the parameters are demonstrated below:

  (a)

  Key complexity The security level offered by the cryptographic algorithm is termed key complexity and is expressed as

  $$\mathrm{key}\mathrm{complexity}=\frac{1}{K}\sum _{a=1}^K\delta (x_a,y_a)$$

  (15)

  Here, the generated $a^{t}h$ key is represented as $y_a$, the identity parameter of $a^{t}h$ key is signified as $x_a$, and K symbolizes the total bit. Moreover, $\delta$ represents the Tanimoto similarity, which is computed by,

  $$\mathrm{key}\mathrm{complexity}=\frac{{\sum }_a\left(max(x_a,y_a)-min(x_a,y_a)\right)}{{\sum }_{a}max(x_a,y_a)}$$

  (16)

  (b)

  Decryption Time

  The decryption time is the time utilized to convert the cipher text into normal plaintext.

  (c)

  Encryption Time

  The time utilized to convert normal plaintext into cipher text is termed encryption time.

• Comparative Model:
  The baseline data-sharing approaches, such as PRE + TEE [3], Medi-Block record [7], MA-RABE [10], and BSKM [12] are utilized to identify the performance of the IABHE+DSNN model designed for secure data-sharing. The comparative analysis was conducted using the Python platform, with the Skin Segmentation database and the Localization Data for Person Activity database. This comparative analysis helps to identify and highlight the improvements in key performance metrics such as key complexity, encryption, and decryption efficiency achieved by our proposed system.

4.2. Comparative Analysis/Validation

The secure data-sharing performance of IABHE+DSNN is validated using the Skin Segmentation database [42] and Localization Data for Person Activity database [43] and is delineated below.

4.2.1. Validation Using Skin Segmentation Database

The analysis of the IABHE+DSNN approach for secure data sharing was conducted using the Skin Segmentation database with data sizes of 100 KB and 200 KB. The detailed results of this analysis are presented below:

• For data size of 100 KB
  The validation of the designed IABHE+DSNN for secure data sharing with varying key sizes using the Skin Segmentation database for 100 KB of data is illustrated in Figure 12. The comparative analysis of key complexity between the proposed IABHE+DSNN and other data-sharing approaches is shown in Figure 12a. Baseline approaches like PRE + TEE, Medi-Block record, MA-RABE, and BSKM measured key complexities of 0.708, 0.747, 0.808, and 0.837, respectively, for a key size of 256 bits, whereas IABHE+DSNN achieved a maximum key complexity of 0.887.
  The investigation of different data-sharing approaches using decryption time is presented in Figure 12b. The IABHE+DSNN recorded a minimum decryption time of 7.987 s, while other traditional approaches recorded decryption times of 20.867 s for PRE + TEE, 17.876 s for Medi-Block record, 12.978 s for MA-RABE, and 10.876 s for BSKM, all for a key size of 256 bits.
  Moreover, the validation of various data-sharing approaches in terms of encryption time is displayed in Figure 12c. The IABHE+DSNN achieved a minimum encryption time of 15.876 s for a key size of 256 bits. In comparison, the encryption times recorded by prevailing approaches were 23.978 s for PRE + TEE, 21.977 s for Medi-Block record, 20.765 s for MA-RABE, and 18.877 s for BSKM.
• For data size of 200 KB
  Figure 13 shows the validation of IABHE+DSNN designed in this research for secure data sharing while utilizing a Segmentation database for 200 KB and by varying key sizes. The comparative analysis of the proposed IABHE+DSNN with other secured data-sharing approaches in terms of key complexity is given in Figure 13a. The IABHE+DSNN obtained a key complexity of 0.575, whereas the baseline secure data-sharing models measured key complexity is 0.398 by PRE + TEE, 0.437 by Medi-Block record, 0.497 by MA-RABE, and 0.536 by BSKM for a key size of 256 Bits. The analysis of different secure data-sharing models using decryption time is given in Figure 13b. The IABHE+DSNN measured minimum decryption time of 17.868 s for a key size of 256 bits, and the decryption time obtained by other prevailing approaches like PRE + TEE is 32.866 s, Medi-Block is 29.867 s, MA-RABE is 25.976 s, and BSKM is 21.978 s. The validation of various data-sharing models utilizing encryption time is given in Figure 13c. The prevailing data-sharing models, such as PRE + TEE, Medi-Block record, MA-RABE, and BSKM, obtained encryption time of 41.786 s, 37.786 s, 33.978 s, and 31.876 s for a key size of 256 Bits, whereas IABHE+DSNN recorded minimum encryption time of 28.876 s.

4.2.2. Validation Using Localization Data for Person Activity Database

The data-sharing performance of IABHE+DSNN is analyzed by utilizing Localization Data for Person Activity database, and the data size varies from 100 KB to 200 KB. The analysis executed is demonstrated below,

• For data size of 100 KB
  The evaluation of IABHE+DSNN used to securely share data while utilizing Localization Data for Person Activity database by differing key sizes for 100 KB data is given in Figure 14. The analysis of various secure data-sharing techniques by employing key complexity is given in Figure 14a. The IABHE+DSNN recorded a maximum key complexity of 0.887, and the key complexity measured for key size of 256 Bits by baseline approaches, such as PRE + TEE is 0.708, Medi-Block is 0.747, MA-RABE is 0.808, and BSKM is 0.837. The comparative validation of data-sharing techniques by means of decryption time is depicted in Figure 14b. The IABHE+DSNN recorded a decryption time of 10.786 s, whereas the decryption time obtained by traditional models for a key size of 256 Bits is 19.876 s by PRE + TEE, 17.876 s by Medi-Block record, 15.876 s by MA-RABE, and 13.735 s by BSKM. In addition, Figure 14c displays the analysis of different data-sharing models by utilizing encryption time. Here, the encryption time recorded by prevailing data-sharing schemes, like PRE + TEE, Medi-Block record, MA-RABE, and BSKM, is 31.876 s, 28.866 s, 22.867 s, and 18.765 s. Similarly, the designed IABHE+DSNN approach obtained an encryption time of 15.765 s, which is less than other existing data-sharing techniques for a key size of 256 bits.
• For data size of 200 KB
  Figure 15 displays the validation of the developed IABHE+DSNN approach with the Localization Data for Person Activity dataset for secure data sharing by varying key sizes. Figure 15a depicts the evaluation of different data-sharing models utilized in the MPA system by means of key complexity. The key complexity obtained by designed IABHE+DSNN is 0.876 for a key size of 256 bits, and the key complexity of 0.708, 0.747, 0.808, and 0.837 is recorded by prevailing secure data-sharing approaches, such as PRE + TEE, Medi-Block record, MA-RABE, and BSKM. Moreover, the validation of data-sharing approaches utilizing decryption time is given in Figure 15b. Here, the decryption time obtained by traditional models for key size of 256 Bits is 27.976 s by PRE + TEE, 22.987 s by Medi-Block record, 19.767 s by MA-RABE, and 17.987 s by BSKM, whereas the designed IABHE+DSNN approach measured minimum decryption time of 13.866 s. The validation of various secure data-sharing schemes using encryption time is displayed in Figure 15c. The encryption time recorded by existing data-sharing techniques, such as PRE + TEE, Medi-Block record, MA-RABE, and BSKM, for key sizes of 256 Bits is 55.987 s, 22.987 s, 19.767 s, and 17.987 s. On the other hand, the IABHE+DSNN technique recorded minimum encryption time of 36.765 s.

4.3. Discussion

The comparative discussion is carried out to identify the performance of IABHE+DSNN in secure data sharing by comparing it with existing data-sharing approaches.

Table 2. Comparison with other approaches.

Size of Data	Evaluation Parameters	PRE + TEE	Medi-Block Record	MA-RABE	BSKM	Proposed IABHE+DSNN
Using skin Segmentation database
100 KB	Key complexity	0.408	0.476	0.545	0.597	0.657
	Decryption time (s)	20.867	17.876	12.978	10.876	7.987
	Encryption time (s)	23.978	21.977	20.765	18.877	15.876
200 KB	Key complexity	0.398	0.437	0.497	0.536	0.575
	Decryption time (s)	32.866	29.867	25.976	21.978	17.867
	Encryption time (s)	41.786	37.786	33.978	31.876	28.876
Using Localization Data for Person Activity database
100 KB	Key complexity	0.708	0.747	0.808	0.837	0.887
	Decryption time (s)	19.876	17.876	15.876	13.765	10.786
	Encryption time (s)	31.876	28.866	22.867	18.765	15.765
200 KB	Key complexity	0.754	0.797	0.826	0.857	0.876
	Decryption time (s)	27.976	22.987	19.767	17.987	13.866
	Encryption time (s)	55.987	49.865	45.987	39.867	36.765

presents the comparative results of the IABHE+DSNN model and existing data-sharing models, with superior performance values highlighted in bold. The minimum decryption and encryption time of 10.786 s and 15.765 s is obtained by designed IABHE+DSNN, and the IABHE+DSNN also recorded maximum key complexity of 0.887 while using Localization Data for Person Activity database and for a key size of 256 Bits. The existing data-sharing techniques, like PRE + TEE, Medi-Block record, MA-RABE, and BSKM obtained encryption time of 31.876 s, 28.866 s, 22.867 s, and 18.765 s, and decryption time of 19.876 s, 17.876 s, 15.876 s, and 13.765 s. Similarly, the key complexity recorded by existing approaches is 0.708 by PRE + TEE, 0.747 by Medi-Block record, 0.808 by MA-RABE, and 0.837 by BSKM. The IABHE+DSNN used for secure data sharing in multi-party data-sharing systems integrates the concepts of data access control and encryption to ensure data security efficiently. Moreover, this model utilizes XOR operations as well as hashing functions to enhance data confidentiality and reduce memory usage in the data-sharing system.

The proposed technique, IABHE+DSNN, excels over existing methods due to its unique combination of Identity and Attribute-Based Honey Encryption Algorithm (IABHE) and Deep Spiking Neural Networks (DSNN). This hybrid approach offers enhanced data confidentiality through identity-based encryption and robust key generation using neural networks. By leveraging the strengths of both encryption and neural network technologies, the proposed technique provides a more secure and efficient solution for data-sharing compared to traditional methods. The integration of IABHE and DSNN results in improved key complexity, reduced decryption and encryption times, and overall higher security levels, making it a superior choice for secure data-sharing applications.

5. Conclusions

This paper introduces a novel deep learning-based key-generation method called IABHE+DSNN for secure data sharing within a blockchain-based multi-party data sharing system. The incorporation of blockchain technology into the IABHE+DSNN method for secure data sharing within a multi-party system provides a robust, transparent, and decentralized framework for managing and sharing sensitive information. The system architecture encompasses essential entities such as the data owner, data requester, IPFS, encryption server, and MPA, ensuring a comprehensive approach to security. Throughout the research, the significance of incorporating robust security functionalities such as hashing, encryption, session passwords, and OTP to safeguard the data-sharing process is emphasized. By following a systematic approach encompassing initialization, initial registration, data protection, authentication, and data-sharing phases, our system ensures the integrity and confidentiality of shared data. Furthermore, the implementation of data protection within the MapReduce framework, leveraging both mapper and reducer phases, demonstrates our commitment to enhancing data security and efficiency. Encryption using IABHE in the mapper phase, coupled with key generation using DSNN, contributes to superior performance metrics, including decryption time, encryption time, and key complexity. The experimental results validate the effectiveness of our proposed IABHE+DSNN approach, demonstrating significant performance improvements over existing techniques. Future research will focus on addressing additional security concerns and analyzing system behavior to further fortify data security within large networks. Our work advances the field of secure data sharing by introducing an innovative approach that combines deep learning-based key generation with blockchain technology, paving the way for enhanced security and confidentiality in multi-party data-sharing environments.