Volume 20 No 20 (2022)
 Download PDF
IMPROVED DEEP REINFORCEMENT LEARNING BASED DEFENSE SYSTEM FOR DISTRIBUTED DENIAL OF SERVICE ATTACK
A. SeemaM.Sc(CS), Dr. P. Shivaranjani M.Sc(CS)
Abstract
Edge computing is the standard integration of cloud and fog computing which has limited access because of the development of Internet-of-Things (IoT) devices and large amount of data are loomed at the edge nodes in the network. Various features of edge computing including on-demand activities, pricing strategies, fast resistance etc., make it more susceptible to different kinds of Distributed Denial of Service (DDoS) attack. It is because of the direct involvement of the edge nodes on the network to enable decisions in real time without a huge amount of information being broadcast by the cloud or edge servers. Therefore, it is more important for identifying and defending these attacks in the network with the help of effective methods. One of the effective methods for defending DDoS attack is Deep Reinforcement Learning (DRL) where the network structure and network traffic information were analyzed using Q-learning to defend the DDoS attack. In DRL, perceptual aliasing occurs when many states share nearly identical features of network structure and traffic information. A consequence of perceptual aliasing is that the DRL agents struggle to learn the relationship between features (network structure and traffic) and utility of particular actions (deployment of classifier in edge server). In order to handle the perceptual aliasing in DRL based DDoS defense system, a Newtonian Action Advice (NAA) is introduced in this paper. In NAA, an advice is provided to take proper action i.e., deploying a classifier in an edge server. Moreover, policy strategy in DRL is improved by engaging combined policy gradient with DRL. It used a value function approximation with the policy which enables to process with continuous action spaces. Based on the Q values, the policy gradient deploys a classifier in the edges for defending DDoS attack.
Keywords
Edge computing, cloud computing, internet-of-things, Distributed Denial of Service, Newtonian Action Advice, combined policy gradient with deep reinforcement learning.
Copyright
Copyright © Neuroquantology

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Articles published in the Neuroquantology are available under Creative Commons Attribution Non-Commercial No Derivatives Licence (CC BY-NC-ND 4.0). Authors retain copyright in their work and grant IJECSE right of first publication under CC BY-NC-ND 4.0. Users have the right to read, download, copy, distribute, print, search, or link to the full texts of articles in this journal, and to use them for any other lawful purpose.