A HYBRID DEEP NEURAL NETWORK BASED CLASSIFICATION METHODS FOR CYBER INTRUSION DETECTION IN SMART CITY ENVIRONMENTS

Abstract

The concept of classification in networking is broadly considered area in computer vision as like that smart city security, management and cyber threat detection is also broadly considered area in smart environments. In Smart cities in Tamilnadu, through the occurrences of excessive things and data flow over networks, increases the cyber-attacks and intrusion detection. Day to day maintaining of this vast amount of network data traffic is very challenging task, consequently DL model based classification algorithms are introduced. Here, MCNN and GRU are proposed to capture the both spatial and temporal features in traffic data can be effortlessly extracted. This proposed model is practically tested on open source BoT-IoT and TON_IoT datasets to provide good results than other methods. Hence, MCGRU model compared with different existing models to produce 99% accurate results in terms of Accuracy, Precision, Recall and F-Score.