Cyber Security Framework to SME Applications using Block Chain Integrated Convolution Neural Network for Authorizing and Classifying Level of Access to Distributed Data
Small and medium size enterprises are becoming critical in driving innovations and economic growth in digital economy. However SME growing reliance on digital technologies exposes to cybersecurity attacks such as data breaches and phishing attacks and ransoms ware attack leads to greater financial loss, reputational challenges and business closure. In order to protect the SME business operation and their process data against cyber security attacks, many researchers applies emerging technologies such as Artificial intelligence and blockchain. Despite of many advantages of the implementing blockchain towards decentralization and transparency while artificial intelligence approaches towards predicting and classifying attacks, it is mandatory to establish an integrated solution to enhance security of the distributed servers of the SME. In this paper, blockchain integrated convolution neural network is designed to predict and classify the user with user level to secure access of datainblockchain enabled distributed servers. Initially Blockchain is established to business process data of the SME with immutable ledger for fostering trust and transparency. Convolution Neural Network establishes access control mechanism to blockchain distributed server to authenticate user against unauthorized access and predict the user level of access to data. In Blockchain, trusted nodes can validate the transaction and request for data access through generation of new transaction by user. User request is logged in blockchain which leads to data transparency and support detect the malicious user to retrieve data in the blockchain. Convolution Neural Network processes the log data of blockchain which contain user request. The user requests were processed in the convolution layer to extract the spatial temporal features. Extracted feature were embedded as spatial embedding and temporal embedding and applied to Max pooling layer. Max pooling layer reduces spatial dimension of the feature map. Spatially reduced feature map is applied to fully connected layer which contains activation function and softmax function to authenticate user and categorize the user with level of access to the data. Experimental analysis of the model is performed in the blockchain platform named as hyperledger which enables convolution neural network for authenticate user and categorize level of user towards data access. Performance analysis of the model proves that model is more secure and accurate against detecting authorized user and classifying user on their level access to data.
Introduction
???? 1. Context and Motivation
SMEs are crucial drivers of innovation, employment, and economic growth.
The adoption of digital technologies (cloud, IoT, AI, blockchain) helps SMEs grow and meet customer expectations.
However, increased digital reliance exposes SMEs to cybersecurity threats, which can lead to data breaches, financial losses, and business failure.
????? 2. Existing Solutions and Challenges
Blockchain improves data integrity and transparency but has complexity in trust establishment and data verification (due to various types like public, private, consortium, and BaaS).
Artificial Intelligence (AI), especially machine learning and deep learning, enhances cybersecurity by detecting complex attack patterns.
Yet, isolated use of AI or blockchain is not sufficient. Integrating both technologies can enhance security, trust, and user verification.
To store business data securely on blockchain and use CNN to predict and classify users by access level for controlled data access.
???? Key Components:
???? Blockchain Layer
A distributed, immutable ledger storing SME business data.
Implements smart contracts and consensus mechanisms (Proof of Authority).
Validates user transactions and shares access links securely.
???? Convolutional Neural Network (CNN)
Extracts spatial features (IP, keys, topology) and temporal features (timestamp, request info) from user requests.
Layers:
Convolution Layer: Feature extraction
Max Pooling Layer: Dimensionality reduction and feature aggregation
Fully Connected Layer: Uses ReLU and Softmax to:
Authenticate user
Classify access level (authorized/unauthorized, and further levels)
???? Integration Flow:
User sends a request with public/private key.
CNN analyzes the request and classifies user.
Validator node uses CNN model to verify and generate block.
Access is granted via an encrypted link; all actions are logged.
???? 4. Experimental Analysis
Conducted using Python (TensorFlow, Keras) with blockchain components coded in Python/JavaScript.
Used CNN pre-trained models and user simulation data (200 users tested).
Dataset from Kaggle includes user transactions, attack scenarios, and access requests.
???? Performance Metrics:
Architecture
Precision (%)
Recall (%)
F-Measure (%)
Blockchain + CNN (Authorized)
96.6
94.2
96.1
Blockchain + CNN (Unauthorized)
96.2
94.5
96.2
Blockchain Only
93.3–93.7
91.5–91.6
93.4–93.8
CNN Only
91.2–92.8
90.2–90.7
91.7
? Overall Accuracy: 96.1%, outperforming standalone blockchain or CNN methods.
???? 5. Security Features
Confidentiality: Transactions are encrypted.
Integrity: User-signed data with private key ensures tamper protection.
Replay Attack Prevention: Use of nonces ensures requests are unique.
Malicious Behavior Detection: Blockchain logs enable monitoring and detection of abnormal access patterns.
???? 6. Contributions and Findings
The blockchain-integrated CNN model offers:
High accuracy in detecting and authorizing users
Dynamic access level classification
Enhanced protection from cyber-attacks
Proves more effective than using blockchain or deep learning models alone.
Conclusion
In this paper, a new blockchain integrated convolution neural network is designed and implemented to secure data access among authorized and unauthorized user along predict and classify the user with user level to secure access of data in blockchain enabled distributed servers. Especially convolution neural network is enabled in blockchain on smart contract and consensus mechanism to validate the user and their access level to the data. On experimental and performance analysis, proposed model is found to be better compared to conventional approaches on securing data access and detecting the authorized user. As a future work, security of the secure access can be enhanced on employing federated architectures.
References
[1] J. Abou Jaoude and R. G. Saade, “Blockchain applications–usage in different domains,” IEEE Access, vol. 7, pp. 45 360–45 381, 2019
[2] F. Casino, T. K. Dasaklis, and C. Patsakis, “A systematic literature review of blockchain-based applications: Current status, classification and open issues,” Telematics and informatics, vol. 36, pp. 55–81, 2019
[3] R. Zou, X. Lv, and J. Zhao, “Spchain: Blockchain- based medical data sharing and privacy-preserving ehealth system,” Information Processing & Man- agement, vol. 58, no. 4, p. 102 604, 2021
[4] M. N. Nobi, R. Krishnan, Y. Huang, M. Shakarami, and R. Sandhu, “Toward deep learning based access control,” in Proceedings of the Twel- veth ACM Conference on Data and Application Security and Privacy, 2022, pp. 143–154
[5] M. N. M. Bhutta, A. A. Khwaja, A. Nadeem, et al., “A survey on blockchain technology: Evo- lution, architecture and security,” IEEE Access, vol. 9, pp. 61 048–61 073, 2021
[6] Guangming Xian\"Cyber Intrusion Prevention for Large-Scale Semi-Supervised Deep Learning Based on Local and Non-Local Regularization\"IEEE access, Vol.8,Page(s): 55526 - 55539, 2020
[7] Ahmed Didouh; Anthony Bahadir Lopez; Houda Labiod; Yassin El Hillali; Atika Rivenq; Mohammad Abdullah Al Faruque\"TileChain: A Blockchain-Based Framework for Security Service Management for Vehicular Communications\" IEEE Transactions on Vehicular Technology, Volume: 74, Issue: 1, pp:1335 - 1351, 2025
[8] Ikram Ud Din; Kamran Habib Khan; Ahmad Almogren; Mahdi Zareei; Jesús Arturo Pérez Díaz\"Securing the Metaverse: A Blockchain-Enabled Zero-Trust Architecture for Virtual Environments\"IEEE Access, Volume: 12, PP: 92337 - 92347,2024
[9] Ozgur Koray Sahingoz; Ebubekir BUBEr; Emin Kugu\"DEPHIDES: Deep Learning Based Phishing Detection System\"IEEE access, Vol.12, PP:8052-8070,2024
[10] B. M. Yakubu, M. I. Khan, A. Khan, F. Jabeen, and G. Jeon, ‘‘Blockchainbased DDoS attack mitigation protocol for device-to-device interaction in smart home,’’ Digit. Commun. Netw., vol. 9, no. 2, pp. 383–392, Apr. 2023.
[11] E. S. Babu, S. Bkn, S. R. Nayak, A. Verma, F. Alqahtani, A. Tolba, and Mukherjee, ‘‘Blockchain-based intrusion detection system of IoT urban data with device authentication against DDoS attacks,’’ Comput. Electr. Eng., vol. 103, Oct. 2022
[12] M. A. Cheema, H. K. Qureshi, C. Chrysostomou, and M. Lestas, ‘‘Utilizing blockchain for distributed machine learning based intrusion detection in Internet of Things,’’ in Proc. 16th Int. Conf. Distrib. Comput. Sensor Syst. (DCOSS), May 2020
[13] T. M. Ghazal, M. K. Hasan, S. N. H. S. Abdullah, K. A. A. Bakar, and H. Al Hamadi, ‘‘Private blockchain-based encryption framework using computational intelligence approach,’’ Egyptian Informat. J., vol. 23, no. 4, pp. 69–75, Dec. 2022
[14] H. N. Abishu, A. M. Seid, Y. H. Yacob, T. Ayall, G. Sun, and G. Liu, ``Consensus mechanism for blockchain-enabled vehicle-to-vehicle energy trading in the internet of electric vehicles,\'\' IEEE Trans. Veh. Technol., vol. 71, no. 1, pp. 946_960, Jan. 2022.
[15] M. Shen, H. Liu, L. Zhu, K. Xu, H. Yu, and X. Du, ``Blockchain-assisted secure device authentication for cross-domain industrial IoT,\'\' IEEE J. Sel. Areas Commun., vol. 38, no. 5, pp. 942_954, May 2020.