• Vinay Kumar
• Abhinav Bhandari

The Internet has grown over the past two decades, and billions of active users and devices has been added to the Internet to make the use of various applications and available data. The rise of Internet has also exaggerated the interest of hackers to disrupt the online services and data. The online services such as financial, defense, etc are considered the highly sensitive datasets, which are always under threat from the online attacks. Also, a number of online services such as social media, email servers, etc, are prone to several forms of attacks due to different reasons. The service disruption attacks are generally launched from a single attacker node or group of nodes, which is known as denial of service (DoS) and distributed denial of service (DDoS) respectively. The proposed model primarily focuses upon the DDoS attack detection, which is considered the most dangerous and effective attacks on resource availability.The proposed model is designed on the basis of machine learning methodology, from which the supervised learning is selected as the primary option to analyze the network data. The supervised learning methods use the training data to learn the possible patterns, and to build the classification related knowledge. The knowledge learnt from the training data is applied to the testing data, which is classified as attacked or normal packet on the basis of its orientation and similarity with the training data. The source and destination IP addresses are decoded in the independent octet values, which convert an IP address to four data point long array. The decoded IP addresses are used to suppress the learning coefficient, and to improvise the quantitative learning instead of non-rooted qualitative learning. The extracted features also include the timestamp, protocol and length, which depict the different aspects of network packets. The ensemble classification, neural network and other machine learning models are deployed on the extracted features to detect DDoS attacks. The Random Forest is found most efficient classification model, whereas KNN & Naïve Bayes classifiers are the second best classifiers in different situations. However, the Multilayer Perceptron (MLP) classifier underperformed on the network data for DDoS detection.

DDoS detection, Intrusion detection, Supervised classification, Ensemble classification

"Ensemble based Classification Model for DDoS Detection using Machine Learning Classifiers", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.5, Issue 8, page no.342-356, August-2018, Available :http://www.jetir.org/papers/JETIR1808529.pdf

"Ensemble based Classification Model for DDoS Detection using Machine Learning Classifiers", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.5, Issue 8, page no. pp342-356, August-2018, Available at : http://www.jetir.org/papers/JETIR1808529.pdf