ESREHO-MaxNet: Deep Maxout Network For Intrusion Detection And Attack Mitigation In Iot With Wrapper Based Feature Selection Approach

An effective intrusion detection method is developed using proposed ESREHO-based Deep Maxout network in the IoT environment. The plant images are captured by the sensor node and are routed to the sink node through CH that is selected by the method named Exponential SFO. The routed data is received at BS, where the intrusion detection strategy is done by undergoing the feature extraction, feature selection, and intrusion detection phase. The log file data generated from the predicted data is fed to feature extraction phase, where the Bot-IoT features are acquired and then the unique features are optimally selected with wrapper model. The Deep Maxout network is employed to detect the intrusions from the data. Then attack mitigation process can be done by reducing the data rate of packets. The proposed method achieves better performance with the measures of accuracy, TPR, energy, and throughput with the values of 0.9418, 0.942, 1.8004J, and 7662438bps for without attack.