Detection and Prevention of Combined Teardrop and Back Attack using SVM and Hybrid CLSTM Algorithm

Abstract

A significant danger to service providers is an Attack through DDoS, or distributed denial of service. This kind of hack bombards the target with too many malicious requests, seeking to interfere and prevent legitimate users from using services. Due to higher operations and financial costs, a DDoS assault causes significant economic loss for enterprises and service providers. Recently, to stop DDoS attacks, numerous methods like IDS and firewall are implemented but they failed to detect many network layer attacks such as teardrop and back attack that makes server unavailable for legitimate users. To overcome the issue this paper uses recent trending approaches like machine learning and deep learning defending mechanism for detection of attacks which can be then integrated to secure servers to protect from combined teardrop and back attack. The study uses and compares two approaches one from machine learning algorithm SVM and other one from hybrid deep learning algorithm CLSTM which is combination of CNN and CLSTM and validates which can perform well in DDOS attack detection. The KDD cup dataset is fed into two approaches the methods concentrate on important feature selection which contributes to attack detection and accurate detection. The ML algorithm SVM provided 73% accuracy and deep learning approach gives 98% accuracy which proves usage of hybrid deep learning is best mechanism for teardrop and back attack detection. For the detected attacks the responsive system is implemented to notify admin with an email and block IPs of attacker system and generate signature for an attack.