Delay Tolerant Network Security: Enhanced Machine Learning Technique for Intrusion Detection System

Delay tolerant networks (DTNs) are intended for effective communication between nodes over huge distances and they are resourceful in extreme conditions. DTN stores and forwards messages when the participant node is in the range. Hence, there is no restriction on end-to-end connectivity and data is transferred over adaptable links between the nodes. Since the network functions on coordination among participant nodes, an untrusted node can affect the coordination. Thus, DTNs are vulnerable to different kinds of attacks influencing the performance of the network. The delay in the transfer of data and unstable connectivity of nodes depends on effective coordination while the possibility of misbehaviour by relay nodes increases network vulnerability to various types of network attacks like packet dropping attacks, flooded attacks, DoS attacks, gray hole attacks, and black hole attack disturbing the network connectivity. Denial of service attacks (DoS) is a major concern in DTN that adversely affects the network. Attacks on DTN can interrupt message delivery and degrade performance. The study proposes the detection of such attacks over DTN with an efficient machine learning (ML) algorithm. The delay tolerant network is a wireless network that transfers information among nodes and is monitored for malicious nodes using a pre-trained ML model. The voting technique is used to enhance the performance of detection. The network attacks are detected with significant accuracy and efficient secure communication is established in the network. Furthermore, the network simulator NS2 is employed to simulate the prevention of malicious attacks in the proposed system. This simulator offers a versatile and customizable environment for modelling various DTN scenarios and assessing the effectiveness of intrusion detection systems in a controlled setting. Our proposed model offers better performance than existing DTN security techniques.