Hybrid deep learning algorithm for smart cities security enhancement through blockchain and internet of things

【Author】 Mishra, Sourav; Chaurasiya, Vijay Kumar

【Source】MULTIMEDIA TOOLS AND APPLICATIONS

【影响因子】2.577

【Abstract】Internet of Things (IoT) usage is cherishing nowadays and it is widely used in variety of exciting application such as hospital, industry and cities. Security is found to be a major concern in IoT based application due to advancement made in the field of Information and software technology. Transactions using internet has become significantly in case of smart cities and it is frequently prone to various cyber security threats. For securing various transaction in smart cities hybrid deep learning LSTM-SVM algorithm is developed. Initially, data related to transaction in smart cities is gathered and pre-processed utilizing min-max normalization and weighted average filter. Min-max normalization is employed or normalizing the raw data to a certain range. Then, weighted average filter is used for removing noise in the data. After that, appropriate features from pre-processed data is selected using reptile search algorithm. Using industrial gateway, the selected features are provided to blockchain based distributed network. Function of blockchain is to verify the transaction process using trusted entities. Verifier check whether the transaction is genuine or fake through detecting the presence of attack. In case of genuine transaction request, the transaction is accomplished through creating a block on the blockchain. On the other hand, if an attack is detected the transaction is blocked and further types of attack is predicted using deep learning-based hybrid LSTM- SVM classifier. Simulation analysis on the proposed hybrid deep learning model showed 97% accuracy, 98% specificity, 91% F1 score and 82% sensitivity. Based on this proposed model fake transaction in IoT based blockchain is blocked and paved way for development of secure smart cities environment.

【Keywords】IoT devices; Min-max normalization; Weighted average filter; Reptile search algorithm; Block chain network; Hybrid LSTM-SVM

【发表时间】2023 2023 AUG 7

【收录时间】2023-08-27

【文献类型】

【主题类别】

--