Wireless Sensor Network-Based Hybrid Intrusion Detection System on Feature Extraction Deep Learning and Reinforcement Learning Techniques

A Wireless Sensor Network (WSN) is one of the most huge parts of the field of correspondence innovation. A Wireless Sensor Network is one kind of remote framework that consolidates endless coursing, self-composed, minute, low controlled contraptions named sensor center points called motes. This innovation has numerous application zones like therapeutic, ecological, transportation, military, amusement, country guard, emergency the board and furthermore keen spaces. Security is one of the most vital aspects concerned with WSN. Intrusion detection (ID) is one of the main issues while concerning about security. This paper is concerned with the comparative study on the existing hybrid intrusion detection method with their advantages in addition to disadvantages. The article also proposes a hybrid interruption recognition system on the basis of feature extraction, deep learning and reinforcement learning techniques which reduces human dependency and takes most decisions automatically. The proposed system integrates anomaly based as well as signature mechanisms for detecting attacks.