Detection of DDOS Attack by Quasi-Newton Back propagation Algorithm

Background and Aim: Distributed Denial of service (DDoS) attacks tries to disrupt a target system and mostly a web server, by flooding it with unauthorized packets, corrupting its bandwidth and overburdening it to prevent legal requests. Distributed Denial of Service (DDoS) attacks used multiple computers and multiple Internet links, an attack that does not make the networks more available to users. The main objective of Distributed Denial of service (DDoS) attacks is to compile different internet-wide systems with infected zombies / botnets of the network. These zombies are intended to attack a specific target or network with different types of packets. This paper presents a compression of two Artificial Neural Networks ۱) feed forward and ۲) Casecade Neural Network. the Quasi-Newton Back propagation Algorithm is used for trainng. both neural network trained and tested with subset of the Knowledge discovery Dataset (KDD) dataset and result checks that the proposed algorithms successfully detects the DDoS attacks and which Network give good accuracy. Methods: ddos attack detection is observeds Results: training algorithms Conclusion: In this research paper feed forward neural network and case cade neural network is used for training detection of DDOS Attack. Matlab ۲۰۱۸a tool is used for this training and result show that the feed forward neural network give good result and take less training time ۲ min ۲۱ sec.in future work different algorithm and neural networks is used to compare that which neural network and algorithms is best for detection of DDoS Attack.

کلیدواژه ها:

Knowledge discovery Dataset (KDD) ، Artificial Neural Network (ANN) ، Distributed Denial of Service (DDoS) Attacks ، Broyden-Fletcher-Goldarb-Shanno(BFGS)