Abstract
A plurality of sensors in a wireless sensor network (W.S.N.) is a physical parameter node that allows detection sensor nodes to interact. Security is an essential issue in wireless sensor networks of many practical applications. Our goal is to launch denial of service attacks and respond to wireless sensor networks to enhance security by detecting the enemy. Different kinds of different layers in the occurrence WSN. These two types of machine learning techniques, neural network (NN), detect a Support Vector Machine (SVM), a media access control (MAC.) layer attacks. I have to compare the two methods. It has an access channel wireless sensor node, MAC. Protective layer is essential. Use scenario probability WSN. Wireless network simulator, Vanderbilt plow error simulation.
1. Introduction
A sensor is a physical article or an assortment of data about a function that happened, sending most sensors for gathering the information remotely to a preparing station. When these sensors are hugely composed of observing the physical climate, they structure a W.S.N. Remote sensor networks speak to a wide assortment of security issues that must be tended. We should consider one of a kind of difficulties. For instance, energy is the primary consideration related to W.S.N. W.S.N. Hubs sensor by a battery or sun oriented force. These are restricted in the information stockpiling asset, processing force and correspondence transfer speed terms.
5. Conclusion
Both SVM and neural network machine learning techniques are used to detect DoS attacks. These are just examples, without re-programming. They are based on supervised learning. Using these methods, malicious nodes can prolong the life of the network to save power. NN. is a distributed parallel systems; a linear program can solve the problem can not be solved. SVM training method using kernel-based to find the global minimum. An analysis of the performance between the two techniques can be found SVM DoS attacks can be detected, accuracy 97%, while the N.N. may reach 91% if SVM is more accurate than NN. It will take a longer time than N.N.; the SVM-based method is more like N.N. Rather than detecting DoS attacks.
