##plugins.themes.bootstrap3.article.main##

K.SOMASUNDARAM

Dr.K.SELVAM

Abstract

In recent times, smart cities have become an effective approach to providing excellent services to the population by efficiently using the resources at hand. While smart cities provide several benefits, ensuring security remains a significant obstacle that has to be addressed. The incorporation of Internet of Things (IoT) applications into the smart city management system, namely in the transportation sector, has the potential to greatly improve urban mobility, efficiency, and sustainability. In this paper, developed Attention-Based Bidirectional Long Short-Term Memory (ABiLSTM) for the efficient and secure data transmission in smart cities IoT based transmission. The proposed ABiLSTM incorporates attention-based scheme integrated with the Bi-LSTM for the data IoT data processing in smart cities. To improve security the ABiLSTM model uses the Advanced weighted AES model for the IoT transportation data encryption and decryption. The encrypted and decrypted data is processed with the Bi-directional LSTM architecture for the efficient and secure data transmission in the smart cities. The implementation is performed with the consideration of real-time IoT with consideration of bus, train and traffic light data in smart cities. Simulation results demonstrated that the ABiLSTM model exhibits a superior performance in terms of encryption and decryption efficiency. Specifically, ABiLSTM achieves an average encryption time of 0.000458 seconds and an average decryption time of 0.000109 seconds. This is notably faster compared to CNN's average encryption time of 0.002888 seconds and decryption time of 0.000327 seconds, and DL's average encryption time of 0.003320 seconds and decryption time of 0.000408 seconds. With the proposed ABiLSTM average data size for encryption is measured as 444bytes which is significantly less than conventional CNN and DL methods achieving 464 and 485 respectively. Additionally, ABiLSTM achieves higher classification performance in terms of accuracy, precision, recall, and F1-Score values of 0.975, 0.950, 0.994 and 0.792 respectively.

##plugins.themes.bootstrap3.article.details##