AN AUTOMATED MULTI-LAYER INTRUSION DETECTION MECHANISM FOR WIRELESS SENSOR NETWORKS
A Mobile Adhoc Network (MANET) is composed of nodes which are spatially distributed over a geographical region. The nodes are autonomous in nature and the network is an infrastructure less network. Each node in the network is free to move in any direction and the links between the nodes will be changed frequently. Each node in the network forwards data packets from other nodes and behaves like a router which receives packets from source nodes and forwards it towards destination nodes. In MANET the security plays a major role and this area has grasped many researchers to propose solutions to various issues and challenges in MANET especially in the area of intrusion detection. This paper focuses on establishing real time intrusion detection in MANET using deep neural networks. The deep neural network is able to classify the network traffic as either normal or abnormal. The neural network based intrusion detection is not limited to a particular type of attack. The model eliminates the need for human effort to write signature of various attacks. The model is trained with historic data to classify the type of abnormality in the traffic such as Gray hole, black hole, forging, packet dropping, and flooding attacks. The accuracy of the model in classifying the network traffic is high when compared to other models designed to address the same problem.
MANET, intrusion detection, deep neural network, network traffic.