SECURING SMART GRIDS WITH DEEP CNNS IN NEXT GENERATION 5G-IOT ECOSYSTEMS

The implementation of 5G networks and IoT devices in smart grid applications facilitates the electricity-generated, distributed, and managed bidirectional transfer of real-time information between utility suppliers and consumers. Nonetheless, this heightened transmission and assurance in IoT devices also introduce unprecedented security issues, as they are susceptible to malicious assaults. Implementing effective attack detection techniques in 5G-IoT smart grid systems for dependable and efficient power distribution, together with the prompt and precise identification of attacks, is essential. A novel technique, termed Target Projection Regressed Gradient Convolutional Neural Network (TPRGCNN), is developed to enhance the accuracy of attack detection in data transmission inside a 5G-IoT smart grid context. The TPRGCNN approach integrates feature selection and classification to enhance secure data transfer by identifying assaults in 5G-IoT smart grid networks. During the feature selection process, TPRGCNN employs the Ruzicka coefficient Dichotonic projection regression approach to improve attack detection accuracy while reducing time complexity. Subsequently, the chosen significant features are input into Jaspen’s correlative stochastic gradient convolutional neural learning classifier for the purpose of attack detection. Classification determines if the transmission is standard or indicative of an assault within the 5G-IoT smart grid network. The implementation findings indicate that the suggested TPRGCNN approach achieves a 5% enhancement in attack detection accuracy and a 2% increase in precision, recall, and F-score, while simultaneously reducing time complexity and space complexity by 13% and 23%, respectively, in comparison to existing methods.