Ensuring security against packet dropping Attacks in wireless adhoc networks

Abstract

There has been an enormous and irreversible shift towards wireless networks in the past ten years. Wireless Adhoc Network is one of the most noteworthy applications among wireless communications. Unlike conventional mobile wireless networks, ad hoc networks are not backed by fixed infrastructure. Instead, each host in the network acts as a router to keep them associated. Self- configuration nature of Adhoc nodes is exploited in military and emergency rescue applications. This unique characteristic of wireless Adhoc networks makes them very appealing to packet-dropping attacks. The proposed research categorizes the packet dropping attacks into three cases: Malicious node attacks, Link spoofing attacks, and Congestion attacks. All three cases are addressed separately, and suitable solutions are proposed for each. To deal with the attacks caused by malicious nodes, an effective solution to detect and prevent the malicious nodes in the network using the Cluster-Based Trust Entropy Method is proposed. Secondly, a Modified AODV with Stochastic Gradient Descent – Deep Learning Neural Network (SGD-DLNN) and Levy Flight – Black Widow Optimization is proposed to deal with congestion attacks. Third, to have lossless packet transmission in the presence of link spoofed message attacks, Elliptic Curve Cryptography Based AODV with enhanced Ant Colony Optimization is proposed. Implementation of proposed algorithms is done in the NS-2 simulator, and performance is evaluated concerning packet delivery ratio, throughput, delay, and routing overhead in the presence and absence of attacks. Moreover, the results significantly improve the network’s overall performance with the proposed approach in all three cases. In addition to the detection, machine learning algorithms are explored to get more accuracy in classifying malicious nodes by trust entropy values.