A HYBRID APPROACH FOR CYBER SECURITY: IMPROVED INTRUSION DETECTION SYSTEM USING ANN-SVM
Abhishek Kajal, Sunil Kumar Nandal
Abstract
The instances of cyber attacks are increasing as of exponential hike in online processing amid COVID-19 pandemic scenario that significantly compromising huge number of confidential data as well. With the rising security demands, the detection of cyber-attacks has emerged as a promising field that offers wider scope for the application developers. In the similar scenario, authors have proposed an improved detection system that could effectively detect attacks as DDoS, malware etc. Initially, Genetic Algorithm is implemented for the feature section and reducing the size of data thereby reducing the computational work load of classifiers. In second phase, Discrete Wavelet Transform (DWT) with Artificial Bee colony (ABC) is used to divides the data into four categories and also filters out the irrelevant features. In later stages, Artificial Neural Network is employed whose results are refined by Support Vector Machine (SVM). The novelty of the work lies in the precise detection of the malicious behaviour of the nodes. This ANN-SVM hybrid approach enhanced the classification efficiency of the proposed system in identifying the cyber-attacks. The simulation study over 1000 rounds reflects the performance of the proposed system in terms of higher precision, recall and f-measure in comparison to the existing works dedicated to deal with DDoS attacks. Also, the proposed work demonstrated better outcomes in terms of average True Positive rate and False Positive rate.