DLDM: Deep learning-based defense mechanism for denial of service attacks in wireless sensor networks

Abstract Wireless Sensor Networks (WSNs) include small battery-based self-governing devices that are deployed in a distributed manner to supervise the environmental or physical circumstances. The routers and gateways are connected to the deployed nodes to support many real-time applications. Due to open access, the security issue arises in WSN. In this circumstance, the external users can be verified by securing authentication is necessary one. In real-time applications, to achieve secured communication they have made many lightweight authentication mechanisms. But WSNs are highly susceptible to DoS attacks as it lacks the synchronization between nodes during data routing. In this paper, a new lightweight DoS detection scheme Deep Learning-based Defense Mechanism (DLDM) has proposed to detect and isolate the attacks in Data Forwarding Phase (DFP). This paper describes the new algorithm for the successful detection of DoS attacks, such as exhaustion, jamming, homing, and flooding. We conduct extensive simulation experiments that can accurately isolate the adversaries and it is more resilient to DoS attacks. Our proposed simulation result shows that it can achieve a high detection rate, throughput, packet delivery ratio, and accuracy. This also reduces the energy consumption and the false alarm rate.