Integrating AI and Blockchain for Enhanced Data Security in IoT-Driven Smart Cities
Abstract
:1. Introduction
- Development of a three-layer operational framework within the IoT comprising sensor, edge, and cloud layers. This structure facilitates precise cyber-threat detection by identifying the transaction abnormalities. The intention is to provide a more accountable and interconnected model of each involved actor during their respective interaction, which needs to be added to the existing modeling approaches;
- Introducing a simplified yet highly robust mechanism for computing the legitimacy score enhances accountability among nodes participating in blockchain transactions. This step addresses issues of a higher degree of complexity involved in trust computation in blockchain operation, yet it cannot offer optimal data privacy;
- Innovation in decentralized Ethereum blockchain operations, integrating AI to optimize data confidentiality, is particularly tailored for smart city applications in the IoT. This approach is intended to mitigate issues about off-chain data storage with increased operational costs;
- Introduction of a simplified consensus-based method and an analytical approach utilizing a decentralized evidence matrix to ensure maximum data integrity, non-repudiation, and confidentiality in large-scale IoT environments. This contributory step is meant to mitigate the possibility of introducing any new attack vector where an attacker can create a fork from a previous block to lower the strength of network security;
- Implementation of a novel metaheuristic optimization-based neural network predictive operation to dynamically identify and classify cyber threats, thereby enhancing system resilience and security. This step addresses the association of the computational burden with most AI-based methods in network security.
2. Related Work
3. Problem Description
4. Methodology
- Edge Layer Operation: This is the first stage of operation in which raw sensory information from smart cities is extracted via multiple gateways and subjected to data confidentiality analysis. The stream of raw sensory information is analyzed to identify the usual traffic patterns with the possibility of a cyber breach. In addition, this monitoring of sensory data also assists in analyzing traffic data to evaluate the management of smart traffic and its responsiveness to malicious activities. This raw sensory data extraction process is carried out within the edge layer to acquire the legitimacy score, where the Ethereum blockchain further initiates validation. The term legitimacy score can be defined as a qualitative measure representing an IoT device’s credibility, integrity, and reliability within a network. It should be noted that this operation stage also involves verifying the legitimacy of all sensing devices. The proposed scheme uses a unique consensus method for preserving privacy by confirming the regular or malicious state of data obtained from various sensors. The edge layer also authenticates the blockchain to safeguard data from multiple cyber threats. The novelty of the proposed architecture lies in its two levels of security approaches associated with data privacy. In the first layer, a newly constructed consensus is used to safeguard the data, whereas the transformation operation is carried out to obtain encoded data from the extracted features. This operation is intended to resist any form of illegal extraction of inference toward attack detection by any malicious adversary. In contrast to conventional blockchain models discussed in the literature, the outcome of blockchain validation is further subjected to transformation using AI, which results in the detection of abnormalities. The analyzed security information is stored in blockchain storage units and forwarded to the next cloud layer via multiple gateway nodes;
- Cloud Layer Operation: Upon arriving at cloud layers via gateway nodes, the data associated with the analyzed security information from the previous layer are initially stored in interconnected and decentralized cluster units of the cloud, where they are further cross-checked for their match with the predefined historical transactional data. This results in malicious/normal activity. It should be noted that two storage mechanisms are involved in the proposed scheme. The first form of storage mechanism is in the edge layer operation, which stores all monitored traffic information and legitimacy scores evaluated on the monitored traffic. The second form of storage mechanism occurs when the outcome of the edge layer detects abnormalities, and this information is stored in various connected storage units involved in the cloud layer operation. The result of this operation is the detection of suspicious activities;
- AI Model: The proposed schemes amend the conventional design of neural networks to address the issues about computational burden reported in the identified research problems in Section 3. The revised version of this neural network model in the proposed AI solution aims to achieve optimized performance in a dynamic environment with faster responsiveness while classifying regular and malicious nodes and traffic. The implemented AI model performs its classification task using the extracted features within the edge layer operation. Finally, the information on analytical operations related to suspicious activity in both the edge and cloud layers is harnessed to confirm the presence of malicious activities. Finally, the operation results in identifying intruders and all patterns of intrusive activities linked to compromising the legitimacy score in the system. The framework offers data confidentiality and results in a highly intellectual system capable of determining any abnormalities within the transactional operation carried out by the blockchain network and cloud environment. Therefore, a two-way security assessment framework was used to validate IoT transactions.
4.1. Evaluation of Legitimacy Score
Algorithm 1. For Evaluating Legitimacy Score. |
Input Output Start 6. 7. Else 8. 9. End 13. 14. Else 15. 16. End End |
4.2. Ethereum Data Reposition
Algorithm 2. For Ethereum Data Reposition. |
Input: Output: Start 2. For 3. extract scores from IoT devices 5. Evaluate data linked to 7. forward(edge, cloud) 8. End 9. End End |
4.3. Ethereum-Based Confidentiality
Algorithm 3. For Ethereum-based Confidentiality. |
Input: (indexed identity of Ethereum block), (prior hash score) Output: (final evidence matrix), (block added) Start ) 4. 5. End 7. 8. End 11. do 12. End 14. ; 16. Else 18. 19. ) 20. End End |
4.4. Analytical Method for Resisting Threats
4.5. AI-Based Security Optimization
5. Results
5.1. Assessment Environment
5.2. Result Accomplishment
5.3. Result Discussion
- Predictive Accuracy: Predictive accuracy is expressed as a percentile and is computed by dividing the number of correct predictions by the total number of predictions. The quantified outcome in Figure 3 shows that the proposed AI model offers approximately 9.3% increased predictive accuracy compared with the mean value of all existing AI models. Lower predictive accuracy scores were observed for KMC (Acc = 82.3%), RNN (Acc = 85.4%), and DT (Acc = 85.2%). The primary reason for this is the unsuitability of these algorithms for capturing highly complex relationships associated with traffic data in a threat environment. However, approaches such as CNN (Acc = 93.7%) and AE (Acc = 92.8%) have better predictive accuracy after the proposed framework. However, the higher computational resource demands for CNN and AE reduce their applicability to yield fewer practical outcomes. The primary reason based on specific scenarios for the optimal predictive accuracy score (Acc = 98.4%) can be explained by two main reasons: (i) the proposed framework performs the analytical operation in both the edge and cloud layers that maintain a higher degree of indexed transactional information to yield better accuracy even before subjecting it to the AI model; (ii) unlike any existing AI models used for secured blockchain operation, the proposed framework implements a neural network integrated with a metaheuristic optimization method acting as multiple levels of faster-operating filtering process to narrow down the predictive outcome to a highly accurate one;
- Processing Time: This performance metric is responsible for computing the overall algorithm processing time for the proposed and conventional AI methods under consideration. Figure 4 shows that the proposed framework reduces the algorithmic processing time by approximately 13% in contrast to the mean values of existing AI models. The justification behind this outcome is as follows. Unlike the existing approaches discussed in Section 2, the proposed framework does not instantly convert the incoming data to the blockchain or directly apply its AI model. This is a particular scenario when the proposed scheme reduces the processing time while the existing system cannot. The proposed framework initially constructs a structure for repositing data over the Ethereum blockchain, followed by building blocks and evaluating the miners. Furthermore, it introduces an analytical method to thwart all possible dynamic threats and then implements the AI model. Hence, the input data to the AI model are characterized by higher-quality data and require a less iterative method for its AI model to generate its predictive outcome. However, this is not the case with conventional AI models, which otherwise apply a series of standalone iterative operations, resulting in a higher processing time. It should also be noted that deep learning-based approaches, such as AE (proctime = 2.9 s), RNN (proctime = 2.5 s), CNN (proctime = 3.8 s), and ANN (proctime = 2.8 s), consume more processing time than conventional machine-learning approaches. It can also be noted that RF (proctime = 0.299 s) offers reduced processing time in contrast to the proposed framework (proctime = 0.389 s); however, it lacks interpretability, and the processing time is expected to increase when exposed to real-time streamed data. It should be noted that the proposed scheme emphasizes processing time (PT) during its final evaluation in contrast to the conventionally adopted metric of Computational Time Complexity (CTC) owing to the following reasons: (i) both PT and CTC are typically used for assessing algorithmic performance; however, CTC provides only a theoretical efficiency measure while PT provides empirical evidence of applying algorithms to practical world scenarios based on hardware; (ii) the CTC parameter emphasizes the asymptomatic behavior of an algorithm by offering high-level performance visualization; however, it abstracts away the inclusion of any implementation specifications while PT offers a concrete measure of the runtime of an algorithm considering extensive attributes, for example, environmental factors, software, hardware, etc.;
- Detection Accuracy: This performance metric was computed by dividing the number of attacks positively detected by the total number of attack test instances introduced. The outcome in Figure 5 shows that the proposed framework offers an improved detection accuracy of approximately 8.1% %in contrast to the mean values of conventional AI models. The proposed AI models offer higher threat detection accuracy, specifically for DDoS attacks (Detacc = 98.5%), Brute-Force attacks (Detacc = 99.5%), spoofing attacks (Detacc = 99.8%), and Recon attacks, mainly concerning vulnerability scan and port scans (Detacc = 98.9%). The proposed framework also showed better threat detection accuracy for Mirai attacks (Detacc = 98.1%). At the same time, there is a less significant difference in detection accuracy performance for identifying host discovery attacks (Detacc = 97.3%) and web-based attacks (Detacc = 97.1%). On the other hand, the existing AI model shows effective threat detection for Mirai attacks (Detacc = 92.4%), Web-based attacks (Detacc = 92.3%), and host discovery attacks (Detacc = 92.1%). Other attacks, such as DDoS, brute-force attacks, spoofing attacks, and Recon attacks, must be more optimally detected by the existing system. It was noted that most of the existing AI models must undergo extensive operation, which is costly and time-consuming, with overfitting issues surfacing. The exact scenario of better performance of the proposed system for higher detection accuracy is noted because of the decentralized blockchain operation performed on multiple edge devices, whose interconnected network is further indexed and hosted in the cloud layer in its distributed storage units. Furthermore, the analytical method implemented in the proposed framework is meant to eliminate unnecessary data and features that reduce computational processing time and offer ample scope for both neural network and metaheuristic optimization. These integrated operations not only allow the system to operate faster to detect any form of abnormalities and inconsistencies but also offer reliable closure toward its inference.
- Transaction Throughput: Higher transaction throughput is always anticipated for any blockchain operation concerning large-scale real-world IoT applications. A closer look at Figure 6 shows that the approaches of Elisa et al. [22], Omar et al. [19], and Qiu et al. [20] recorded the lowest transaction throughputs. Elisa et al. [22] introduced an authentication mechanism that considers multiple contents of blockchain addresses, such as user identity, transactions, and record numbers. At the same time, these values continue to escalate, demanding more computational effort toward iteratively reforming authentication. Hence, the throughput declines. Omar et al. [19] offered a segregated structure for system applications with intrinsic data, while the blockchain was kept as an external structure. Hence, fetching services and verifying many users are witnessed with a reduced throughput. Qiu et al. [20] introduced a private blockchain with sophisticated query request processing using dynamic location variables. The anonymizer module performs its task effectively to obfuscate the query for the server. However, when exposed to a dynamic threat environment, this approach demands a higher dependency on the blockchain network to undergo a re-analysis process concerning its query. Hence, the throughput drops significantly, even though it is one of the best static data/transaction approaches. However, the approaches of Ullah et al. [25], Yousra et al. [27], and Aguero et al. [16] have been shown to offer better throughput after the proposed framework. However, these frameworks are specifically designed for particular applications that need to be more flexible in supporting generalized IoT applications with a more significant stream of transactions. Two factors of the explicit scenario can justify the optimal throughput results for the proposed model: (i) the mechanism of evaluation and assessment of the legitimacy score offers more accountable nodes to participate in the transaction process, while the new Ethereum design generates a final evidence matrix that reduces the computational effort required by a system hosted in edge devices, and (ii) a novel neural network-based approach with the inclusion of dynamic weight and bias tuning with selection of optimal conditions leads to more accountable records needed to support a large number of transactions. The quantified outcome shows that the proposed framework offers an approximately 20% increase in throughput compared with the mean scores of the blockchain-based approaches;
- Resource Consumption: Almost every blockchain operation includes extensive computational resources, and consistency increases with more users joining the network. A closer look at Figure 7 shows that the proposed framework offers significantly lower resource consumption. In contrast, the approaches of Lee and Song [18], Ugochukwu et al. [24], and Aguero et al. [16] offer significantly higher resource consumption. The approach proposed by Lee and Song [18] was used to deploy ring signatures to develop a blockchain structure. Although this adoption offers better privacy preservation by hiding the sender’s and receiver’s addresses, its smart contract method extensively deploys symmetric key encryption, which increases the dependencies of secret key storage within the nodes. This architecture was initially designed for the healthcare sector in IoT; however, when exposed to a much larger-scale IoT environment with multiple constructed application domains, the resource dependencies significantly increase. The study model implemented by Ugochukwu et al. [24] suffered from similar challenges associated with smart contract operations between IoT devices and blockchain networks. The framework presented by Aguero et al. [16] involves many software components to manage node identification, including the cyclic process of managing and retaining identifiers. This cyclic task offers a significant hurdle to the intruder; however, it also impedes an average user, preventing them from undergoing similar authentication iteratively. Although this study model provides genuine resistance to multiple levels of security threats, there is a trade-off between data integrity and data non-repudiation when this model is exposed to a dynamic form of cyber threats in a large environment. The prime event when the proposed system is found to excel at optimal performance in contrast to the existing scheme can be justified by three factors concerning the reduced score of resource consumption for the proposed framework: (i) the proposed framework progressively increases the data quality in every incremental step of operation, leading to less computational effort toward the minimized size of data; (ii) the encoding mechanism further compliments this evaluation process during verification with much less computational effort and data dependencies; and (iii) the method of generating evidence in the consensus approach and hash-based integrity checks is carried out using extracted features and not raw data, leading to increasingly lower resource dependencies. The quantified outcome shows that the proposed framework offers approximately 22% reduced resource consumption compared to existing AI models;
- Confirmation Time: This performance parameter represents the system’s response time toward validating transactions, which is essential for any blockchain-deployed IoT application that demands faster responsiveness. This responsiveness depends on the structure of the blockchain and its integration into the system. Figure 8 shows the existing approaches to encounter slightly longer confirmation times. Although this higher confirmation time score is no more than 4 s, they can go extensively with many transactions on streamed IoT applications. Ullah et al. [25] reported a higher confirmation time of 3.887 s when analyzed in a standard test environment in the proposed analysis. The Merkle root tree has been used to manage use-case data, increasing the computational overhead due to data block hashing and integration. This increased the confirmation time. The blockchain model presented by Aguero et al. [16] was also found to have a higher confirmation time of 3.107 s, which is mainly due to the involvement of an external transaction manager in validating the account. This mechanism includes extensive validation and unlocking of information using sophisticated passphrase management. Although this framework offers better data integrity and a higher degree of bidirectional secrecy in cryptography, its extensive operation requires more resources and validation time. The model presented by Aldyaflah et al. [21] also exhibited a slightly longer confirmation time of 2.544 s. This model introduced an access control system using the roles of users for better data confidentiality, whereas smart contracts were used as secured data stores. The data structure used involves extensive mapping of tag indexes with the database, offering better data secrecy; however, fetching and query management for concurrent clients on a large scale is challenging, apart from including a higher confirmation time. However, the reduced confirmation time for the proposed framework was mainly attributed to the inclusion of similarity measures for the optimal selection of features. Furthermore, the encoding process performed on features offers extensive resistance to dynamic cyber threats and a lightweight transformation process. This phenomenon within the proposed scheme is another specific scenario that reduces the dependency on iterative validation, even on a large scale, and for concurrent users, reducing confirmation time. Unlike existing blockchain models, the quantified outcome shows that the proposed model offers an approximately 19% reduced confirmation time.
6. Conclusions
Author Contributions
Funding
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- Anwar, F.; Khan, B.U.I.; Kiah, M.L.B.M.; Abdullah, N.A.; Goh, K.W. A Comprehensive Insight into Blockchain Technology: Past Development, Present Impact and Future Considerations. Int. J. Adv. Comput. Sci. Appl. 2022, 13, 878–907. [Google Scholar] [CrossRef]
- Al Hwaitat, A.K.; Almaiah, M.A.; Ali, A.; Al-Otaibi, S.; Shishakly, R.; Lutfi, A.; Alrawad, M. A new blockchain-based authentication framework for secure IoT networks. Electronics 2023, 12, 3618. [Google Scholar] [CrossRef]
- Khan, B.U.I.; Goh, K.W.; Mir, M.S.; Mohd Rosely, N.F.L.; Mir, A.A.; Chaimanee, M. Blockchain-Enhanced Sensor-as-a-Service (SEaaS) in IoT: Leveraging Blockchain for Efficient and Secure Sensing Data Transactions. Information 2024, 15, 212. [Google Scholar] [CrossRef]
- Xu, D.; Gao, Y.; Xiao, X. Precision Poverty Alleviation Methods in the Agricultural Field Based Upon Wireless Communication Networks and Blockchain. Wirel. Commun. Mob. Comput. 2022, 2022, 2687445. [Google Scholar] [CrossRef]
- Mathur, S.; Kalla, A.; Gür, G.; Bohra, M.K.; Liyanage, M. A Survey on Role of Blockchain for IoT: Applications and Technical Aspects. Comput. Netw. 2023, 227, 109726. [Google Scholar] [CrossRef]
- Abubakar, M.; Jarocheh, Z.; Al-Dubai, A.; Liu, X. A Survey on the Integration of Blockchain and IoT: Challenges and Opportunities. In Advanced Sciences and Technologies for Security Applications; Springer International Publishing: Cham, Switzerland, 2022; pp. 197–221. ISBN 9783031044236. [Google Scholar]
- Alam, T. Blockchain-Based Internet of Things: Review, Current Trends, Applications, and Future Challenges. Computers 2022, 12, 6. [Google Scholar] [CrossRef]
- Sathish, C.; Rubavathi, C.Y. A Survey on Blockchain Mechanisms (BCM) Based on Internet of Things (IoT) Applications. Multimed. Tools Appl. 2022, 81, 33419–33458. [Google Scholar] [CrossRef]
- Imran, M.; Zaman, U.; Imran; Imtiaz, J.; Fayaz, M.; Gwak, J. Comprehensive Survey of IoT, Machine Learning, and Blockchain for Health Care Applications: A Topical Assessment for Pandemic Preparedness, Challenges, and Solutions. Electronics 2021, 10, 2501. [Google Scholar] [CrossRef]
- Hariharan, R.; Tyagi, A.K.; Soni, G. A Survey on Blockchain-Internet of Things-Based Solutions. In Privacy Preservation and Secured Data Storage in Cloud Computing; IGI Global: Hershey, PA, USA, 2023; pp. 108–134. ISBN 9798369305935. [Google Scholar]
- Alwi, S.; Salleh, M.M.; Abu, M.; Ismail, A.F.; Abbas, M.S.; Fadzilah, A.H.H. Concept of Integration of Blockchain and Artificial Intelligence. In Proceedings of the 2023 3rd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE), Greater Noida, India, 12–13 May 2023. [Google Scholar] [CrossRef]
- Chithanuru, V.; Ramaiah, M. An Anomaly Detection on Blockchain Infrastructure Using Artificial Intelligence Techniques: Challenges and Future Directions—A Review. Concurr. Comput. 2023, 35, e7724. [Google Scholar] [CrossRef]
- Soori, M.; Dastres, R.; Arezoo, B. AI-Powered Blockchain Technology in Industry 4.0, A Review. J. Econ. Technol. 2023, 1, 222–241. [Google Scholar] [CrossRef]
- Atlam, H.F.; Azad, M.A.; Alzahrani, A.G.; Wills, G. A Review of Blockchain in Internet of Things and AI. Big Data Cogn. Comput. 2020, 4, 28. [Google Scholar] [CrossRef]
- Ouyang, L.; Zhang, W.; Wang, F.-Y. Intelligent Contracts: Making Smart Contracts Smart for Blockchain Intelligence. Comput. Electr. Eng. 2022, 104, 108421. [Google Scholar] [CrossRef]
- Gutierrez-Aguero, I.; Anguita, S.; Larrucea, X.; Gomez-Goiri, A.; Urquizu, B. Burnable Pseudo-Identity: A Non-Binding Anonymous Identity Method for Ethereum. IEEE Access 2021, 9, 108912–108923. [Google Scholar] [CrossRef]
- Javed, I.T.; Alharbi, F.; Margaria, T.; Crespi, N.; Qureshi, K.N. PETchain: A Blockchain-Based Privacy Enhancing Technology. IEEE Access 2021, 9, 41129–41143. [Google Scholar] [CrossRef]
- Lee, D.; Song, M. MEXchange: A Privacy-Preserving Blockchain-Based Framework for Health Information Exchange Using Ring Signature and Stealth Address. IEEE Access 2021, 9, 158122–158139. [Google Scholar] [CrossRef]
- Omar, A.A.; Jamil, A.K.; Khandakar, A.; Uzzal, A.R.; Bosri, R.; Mansoor, N.; Rahman, M.S. A Transparent and Privacy-Preserving Healthcare Platform with Novel Smart Contract for Smart Cities. IEEE Access 2021, 9, 90738–90749. [Google Scholar] [CrossRef]
- Qiu, Y.; Liu, Y.; Li, X.; Chen, J. A Novel Location Privacy-Preserving Approach Based on Blockchain. Sensors 2020, 20, 3519. [Google Scholar] [CrossRef]
- Aldyaflah, I.M.; Zhao, W.; Upadhyay, H.; Lagos, L. The Design and Implementation of a Secure Datastore Based on Ethereum Smart Contract. Appl. Sci. 2023, 13, 5282. [Google Scholar] [CrossRef]
- Elisa, N.; Yang, L.; Chao, F.; Naik, N.; Boongoen, T. A Secure and Privacy-Preserving E-Government Framework Using Blockchain and Artificial Immunity. IEEE Access 2023, 11, 8773–8789. [Google Scholar] [CrossRef]
- Khor, J.H.; Sidorov, M.; Zulqarnain, S.A.B. Scalable Lightweight Protocol for Interoperable Public Blockchain-Based Supply Chain Ownership Management. Sensors 2023, 23, 3433. [Google Scholar] [CrossRef]
- Ugochukwu, N.A.; Goyal, S.B.; Rajawat, A.S.; Verma, C.; Illés, Z. Enhancing Logistics with the Internet of Things: A Secured and Efficient Distribution and Storage Model Utilizing Blockchain Innovations and Interplanetary File System. IEEE Access 2024, 12, 4139–4152. [Google Scholar] [CrossRef]
- Ullah, A.; Siddiquee, S.M.S.; Hossain, M.A.; Ray, S.K. An Ethereum Blockchain-Based Prototype for Data Security of Regulated Electricity Market. Inventions 2020, 5, 58. [Google Scholar] [CrossRef]
- Viswanadham, Y.V.R.S.; Jayavel, K. A Framework for Data Privacy Preserving in Supply Chain Management Using Hybrid Meta-Heuristic Algorithm with Ethereum Blockchain Technology. Electronics 2023, 12, 1404. [Google Scholar] [CrossRef]
- Yousra, B.; Yassine, S.; Yassine, M.; Said, S.; Lo’ai, T.; Salah, K. A Novel Secure and Privacy-Preserving Model for OpenID Connect Based on Blockchain. IEEE Access 2023, 11, 67660–67678. [Google Scholar] [CrossRef]
- Stefanescu, D.; Montalvillo, L.; Galán-García, P.; Unzilla, J.; Urbieta, A. Industry 4.0 Business-Oriented Blockchain Design Decision Tree. In Blockchain and Applications, 5th International Congress; Springer Nature: Cham, Switzerland, 2023; pp. 113–123. ISBN 9783031451546. [Google Scholar]
- Fu, M.; Zhang, C.; Hu, C.; Wu, T.; Dong, J.; Zhu, L. Achieving Verifiable Decision Tree Prediction on Hybrid Blockchains. Entropy 2023, 25, 1058. [Google Scholar] [CrossRef]
- Salb, M.; Zivkovic, M.; Bacanin, N.; Chhabra, A.; Suresh, M. Support Vector Machine Performance Improvements for Cryptocurrency Value Forecasting by Enhanced Sine Cosine Algorithm. In Computer Vision and Robotics; Springer: Singapore, 2022; pp. 527–536. ISBN 9789811682247. [Google Scholar]
- Monteiro, S.; Oliveira, D.; António, J.; Henriques, J.; Martins, P.; Wanzeller, C.; Caldeira, F. A Scalable Framework to Predict Bitcoin Price Using Support Vector Machine. In Advances in Intelligent Systems and Computing; Springer International Publishing: Cham, Switzerland, 2023; pp. 293–299. ISBN 9783031148583. [Google Scholar]
- Inder, S.; Sharma, S. Predicting the Movement of Cryptocurrency “Bitcoin” Using Random Forest. In Communications in Computer and Information Science; Springer International Publishing: Cham, Switzerland, 2021; pp. 166–180. ISBN 9783030912437. [Google Scholar]
- Ivaninskiy, I.; Ivashkovskaya, I. Are Blockchain-Based Digital Transformation and Ecosystem-Based Business Models Mutually Reinforcing? The Principal-Agent Conflict Perspective. Eurasian Bus. Rev. 2022, 12, 643–670. [Google Scholar] [CrossRef]
- Lawrence, T.; Zhang, L. IoTNet: An Efficient and Accurate Convolutional Neural Network for IoT Devices. Sensors 2019, 19, 5541. [Google Scholar] [CrossRef]
- Uppala, R.; Ramya, V.; Senthil, M.V. Develop a 7 Layers Convolution Neural Network and IoT-Based Garbage Classification System. Int. J. Intell. Syst. Appl. Eng. 2023, 11, 268–275. [Google Scholar]
- Kim, S.-K.; Huh, J.-H. Artificial Neural Network Blockchain Techniques for Healthcare System: Focusing on the Personal Health Records. Electronics 2020, 9, 763. [Google Scholar] [CrossRef]
- HaddadPajouh, H.; Dehghantanha, A.; Khayami, R.; Choo, K.-K.R. A Deep Recurrent Neural Network Based Approach for Internet of Things Malware Threat Hunting. Future Gener. Comput. Syst. 2018, 85, 88–96. [Google Scholar] [CrossRef]
- Alamro, H.; Marzouk, R.; Alruwais, N.; Negm, N.; Aljameel, S.S.; Khalid, M.; Hamza, M.A.; Alsaid, M.I. Modeling of Blockchain Assisted Intrusion Detection on IoT Healthcare System Using Ant Lion Optimizer with Hybrid Deep Learning. IEEE Access 2023, 11, 82199–82207. [Google Scholar] [CrossRef]
- Arifeen, M.; Ghosh, T.; Islam, R.; Ashiquzzaman, A.; Yoon, J.; Kim, J. Autoencoder Based Consensus Mechanism for Blockchain-Enabled Industrial Internet of Things. Internet Things 2022, 19, 100575. [Google Scholar] [CrossRef]
- Alaghbari, K.A.; Lim, H.-S.; Saad, M.H.M.; Yong, Y.S. Deep Autoencoder-Based Integrated Model for Anomaly Detection and Efficient Feature Extraction in IoT Networks. IoT 2023, 4, 345–365. [Google Scholar] [CrossRef]
- Taher, S.S.H.; Ameen, S.Y.; Ahmed, J.A. Enhancing Blockchain Scalability with Snake Optimization Algorithm: A Novel Approach. Front. Blockchain 2024, 7. [Google Scholar] [CrossRef]
- Singh, R.; Ujjwal, R.L. Hybridized Bio-Inspired Intrusion Detection System for Internet of Things. Front. Big Data 2023, 6, 1081466. [Google Scholar] [CrossRef]
- Sangeeta, N.; Nam, S.Y. Blockchain and Interplanetary File System (IPFS)-Based Data Storage System for Vehicular Networks with Keyword Search Capability. Electronics 2023, 12, 1545. [Google Scholar] [CrossRef]
- Olanrewaju, R.F.; Khan, B.U.I.; Kiah, M.L.M.; Abdullah, N.A.; Goh, K.W. Decentralized Blockchain Network for Resisting Side-Channel Attacks in Mobility-Based IoT. Electronics 2022, 11, 3982. [Google Scholar] [CrossRef]
- Ethereum Ethereum/Go-Ethereum: Official GO Implementation of the Ethereum Protocol, GitHub. Available online: https://github.com/ethereum/go-ethereum (accessed on 3 February 2024).
- Etherscan APIs-Ethereum (ETH) API Provider. Available online: https://etherscan.io/apis (accessed on 14 February 2024).
- CIC IoT Dataset 2023. Available online: https://www.unb.ca/cic/datasets/iotdataset-2023.html (accessed on 8 February 2024).
Parameter | Values |
---|---|
Number of accounts | 200 |
Number of transactions in 1 batch | 20 |
Number of peers/devices | 10 |
Type of genesis block | override |
Generated transaction rate | 0.01 s |
Frequencies of transaction | 40,000 |
IoT devices | 200 |
Mode of smart contract | Independent execution |
Approaches | Accuracy (%) | Algorithm Processing Time (s) |
---|---|---|
Proposed | 98.4 | 0.389 |
Decision Tree (DT) | 85.2 | 0.687 |
Support Vector Machine (SVM) | 87.5 | 0.509 |
Random Forest (RF) | 90.5 | 0.299 |
Logistic Regression (LR) | 88.7 | 0.898 |
K-Means Clustering (KMC) | 82.3 | 1.207 |
Reinforcement Learning (RL) | 92.2 | 0.908 |
Artificial Neural Network (ANN) | 91.5 | 2.871 |
Convolution Neural Network (CNN) | 93.7 | 3.803 |
Recurrent Neural Network (RNN) | 85.4 | 2.596 |
Long Short-Term Memory (LSTM) | 90.1 | 1.977 |
Auto Encoder (AE) | 92.8 | 2.902 |
Classes of Adversary | Proposed | Existing System |
---|---|---|
DDoS | 98.5 | 90.4 |
Brute-Force | 99.5 | 89.6 |
Spoofing | 99.8 | 87.8 |
Recon | 98.9 | 87.9 |
Host Discovery | 97.3 | 92.1 |
Web-based | 97.1 | 92.3 |
Mirai | 98.1 | 92.4 |
Blockchain Approaches | Transaction Throughput | Resource Consumption | Confirmation Time | Detection Accuracy | Processing Time |
---|---|---|---|---|---|
Proposed | 4500 | 29.87 | 0.2665 | 98.65 | 0.6766 |
Aldyaflah et al [21] | 2765 | 49.89 | 2.544 | 91.45 | 5.9978 |
Elisa et al [22] | 1988 | 47.83 | 1.006 | 89.03 | 1.1886 |
Javed et al. [17] | 2296 | 42.11 | 2.313 | 89.57 | 5.872 |
Khor et al. [23] | 2011 | 47.29 | 1.926 | 87.06 | 2.6088 |
Lee and Song [18] | 2981 | 62.56 | 3.216 | 90.02 | 6.446 |
Omar et al. [19] | 1989 | 51.99 | 1.132 | 85.11 | 3.897 |
Qiu et al. [20] | 1303 | 51.05 | 0.997 | 87.11 | 2.651 |
Ugochukwu et al. [24] | 3101 | 62.12 | 1.093 | 91.37 | 2.196 |
Ullah et al. [25] | 3655 | 54.13 | 3.887 | 91.76 | 4.127 |
Viswanadham and Jayavel [26] | 2199 | 55.36 | 2.187 | 90.1 | 4.302 |
Yousra et al. [27] | 3211 | 43.87 | 2.876 | 92.67 | 4.968 |
Aguero et al. [16] | 3266 | 65.02 | 3.107 | 91.52 | 5.302 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Khan, B.U.I.; Goh, K.W.; Khan, A.R.; Zuhairi, M.F.; Chaimanee, M. Integrating AI and Blockchain for Enhanced Data Security in IoT-Driven Smart Cities. Processes 2024, 12, 1825. https://doi.org/10.3390/pr12091825
Khan BUI, Goh KW, Khan AR, Zuhairi MF, Chaimanee M. Integrating AI and Blockchain for Enhanced Data Security in IoT-Driven Smart Cities. Processes. 2024; 12(9):1825. https://doi.org/10.3390/pr12091825
Chicago/Turabian StyleKhan, Burhan Ul Islam, Khang Wen Goh, Abdul Raouf Khan, Megat F. Zuhairi, and Mesith Chaimanee. 2024. "Integrating AI and Blockchain for Enhanced Data Security in IoT-Driven Smart Cities" Processes 12, no. 9: 1825. https://doi.org/10.3390/pr12091825