Detection of Spoofed IP nodes using BAT Algorithm and Extreme Learning Machine

Abstract

IP spoofing is known as the most important cyber-attack which is the source for DoS or DDoS attacks where the attacker is hidden inside the network and makes the computer resource services unavailable to the users. The attacker once done with spoofing the IP address will start to flood the system with keeping on sending requests and make the network bandwidth slow to the extent. This paper contains the literature study of the different types of defence mechanisms from different authors used few decades before to detect and mitigate the Spoofed IP nodes at router, host level and recently some author come up with ideas of using computational intelligence methods for detecting the different types of attacks in wireless communications which results in accurate prediction. This paper provides creating a threat model of detecting the Spoofed IP nodes among 105 network wireless communication scenario using computational intelligence algorithm, the features are selected from the simulated raw data and preprocessed by using BAT optimization algorithm and features are converted to ELM readable format and then they are trained and learned using Extreme learning machine algorithm to predict the accurate detection of the Spoofed IP nodes in the wireless communication network scenario. The proposed method provides high accuracy in detection of Spoofed IP nodes with respect to some performance metrics like end to end delay, throughput, packet delivery ratio, packet drop ratio and it is compared with the KNN-SVM exiting model proved the results.