An Integrated Approach for detecting DDoS attacks in Cloud Computing

Abstract

With the recent advancement of Cloud Computing, it provides various services for both organizational and individual users such as shared computing resources, storage, networking etc on demand. Even though it offers various benefits to the users still it remains exposed to many types of attacks which attract cyber criminals. The most common type of attacks on Cloud computing is Distributed Denial of Service (DDoS) Attack. DDoS attack is an attack which makes resources unavailable to the user by compromising large number of systems called bots. The attacker infects various systems in order to carry the attack called Botnet. This thesis aims to implement detection of DDoS attacks through Feature based selection algorithms. For this NSL KDD dataset is used which is a benchmark for Network Intrusion and Detection systems. Feature Selection also called Variable selection or Attribute selection is the method of choosing a subset of significant features for constructing a model. NSL KDD consists of 41 features and categorised as either normal or attack. The attacks are divided into 4 categories: DoS, Probe, U2R and R2L. We divide the dataset into training set and testing datasets by applying 10 fold cross validation. Based on the results we apply classification algorithms such as Decision trees, Random Forest, KNN classification, Naïve Bayes classifier etc and evaluate its accuracy. We can then evaluate which algorithm is better and how it can be better compared to other algorithms.