Deep Learning Assisted DoS Attack Prevention in Wireless Sensor Networks

Abstract

The Wireless Sensor Networks (WSN) are more vulnerable to the Denial of Service (DOS) attacks which disrupts the entire operations of the network. The Jamming attacks in WSN is a DOS attack which blocks either single or cluster of nodes in communicating and denies the services to the indorsed nodes. However, the extenuation of Jamming attacks during routing is a difficult task. In order to mitigate these attacks in WSN an intelligence mitigation technique using Generative Adversarial Network (GAN) method combined with RNN (Recurrent Neural Network) is proposed, to route the data packets between the source and destination nodes by finding the optimal paths. This mitigation technique uses the aid of machine learning to detect the Jamming attack in WSN. The performance of the system is tested against various network parameters that includes control packets overhead, packet delivery rate, energy consumption, network lifetime, and throughput by using COOJA simulator. The results of simulation show that the proposed method obtains improved prevention of various attacks under machine learning assisted framework than the other conventional methods.