Defending DDoS Attack using Stochastic Model based Puzzle Controller

Distributed denial of service (DDoS) attack aimed at network is some sort of malicious activity or unusual behavior, which compromise the availability of the server's resources and prevents the legitimate users from using the service. In this paper a preventive mechanism using stochastic model based puzzle controller, to eliminate the possibility of DDoS attack is proposed. A behavior matrix is defined to record the behavior of users in terms of number of requests sent. A puzzle controller based on stochastic model is used to analyze the behavior matrix and compute covariance matrix. The entropy computed from the covariance matrix is compared against the threshold to detect DDoS attack. To defend against DDoS attack, allocate resources only to those clients who solve the puzzle with difficulty level determined based on entropy value.