Prevention Mechanism on DDOS Attacks by using Multilevel Filtering of Distributed Firewalls

Abstract

Abstract-In the past decade, it has been found that DDoS has proved to be the most dangerous attack. IP spoofing is one of the kinds of DDoS attack which is emerging as a big threat in today's world of technology. The Proposed Framework is a unique technique composed of distributed firewalls and hop count based filtering, it can be used to prevent such kind of DDoS attack. At The primary level, distributed firewalls filters the IP addresses which can be either Internet or Intranet. Along with this it also provide various other advantages such as reducing the dependency of network topology. At secondary level, hop count and TTL based filtering provides more secure level of filtration. In this paper, we have proposed a new framework which reduces the limitations of previous conventional techniques.

Explainable Deep Neural Network Based Analysis on Intrusion Detection Systems

“…This concept of a methodology with locally concentrated and globally concentrated interpretations in the same instance has strong theoretical and mathematical support when compared with other methodologies. The concept of SHAP [24] links the concept of LIME [21] and Shapley values [14]. LIME (local interpretable model-agnostic explanation) [21] concentrates more on learning a local replacement model in order to evaluate its individual forecasts.…”

“…( 1), the LIME model trains interpretable local surrogate framework j on the newly obtained data set by decrementing the loss function and then explores the prediction of a sample by interpreting local framework Ψ(a). Shapley explained the evaluation methodology of obtaining Shapley values [14]; this methodology is utilized in game theory to ascertain the proportion of each individual of a game. This process can be more understandable by utilizing the concepts that are related to the predictions of machine-learning methodologies.…”