DEEP LEARNING ALGORITHM FOR THE DETECTION OF BOTNET ATTACK IN CLOUD ENVIRONMENT

Abstract

ABSTRACT The upsurge in the use of computer, internet, and cloud-based resources and vectors in everyday dealings for a lot of businesses or general communication due to its costeffectiveness and efficiency has made cloud services vulnerable to attacks including DDoS. Botnets also called Bots is the number of Internet-connected devices, which is running one or more bots each. Botnets can be used to perform distributed denial-of-service attack, steal data, send spam, and allows attackers to access devices and their connections. In this study, a proposed Deep Neural Network learning algorithm for cloud botnet attack is presented which comprises of 5 layers architecture. In which the first layer is the input, the second, third and fourth layers are the hidden layer, while the fifth layer represents the output later. In the input layer 42 neurons is composed which equate to the number of features found in the adopted dataset. The processing of the input data for the model building takes place in hidden layer through weights manipulation, the input data was transformed based on the non-linear function known as activation function. The activation function adopted for this research are rectified linear unit (RELU) for the hidden layers and sigmoid function for the output layer. The performance of the algorithm was measured in terms of accuracy, sensitivity, false positive rate, false alarm rate and detection rate respectively. The achievement recorded from the proposed DNN learning algorithm model indicates that an accuracy of 99.65% was achieved which reflects to the high strength of the proposed model in distinguishing a cloud botnet attack from a normal internet traffic flows, in the same manner sensitivity score of 99.68% was achieved from the proposed model while 3.5% and 0.4% was achieved for false alarm rate (FAR) and False positive rate (FPR) respectively