Protecting the WSN by Detecting and Disabling the Affected Sensor Nodes Using NN and SVM Approaches

Abstract

Maintaining Wireless Sensor Network (WSN) security is one of the most important ways to monitor real-time systems in general.Denial-of-Service (DoS) attack is one of the most important things that interfere with the safety of WSN; this is because the attack suddenly shut down the network. Moreover, this attack can mislead the results of the system. So this paper uses two types of techniques to detect attack in MAC layer.That is, it handles two of the most advanced systems in machine learning. One of them is the Neural Network (NN) and the other is the Support Vector Machine (SVM) method. In simulation, the parameters which are critical and normalized are calculated as well probabilities of the associated Denial of Service attack calculated in the test runs.The above calculated values are used as inputs for training the system in SVM and NN. In the simulation results, accuracy of the result by using SVM is more accurate compared to NN's method.Both Support Vector Machine and Neural Network methods are very useful in determining the percentage of possibility of DoS attacks in Wireless Sensor Network.