Author(s): Usha Maheshwari, A Varshitha, D Sri Gouthami and K Sai Sirisha

Abstract: This study introduces an AI-based security component to an Internet of Things (IoT) paradigm based on Mobile Ad Hoc Networks (MANETs). When an attacker node in a MANET suddenly stops receiving any data at all, a phenomenon known as a Black Hole Attack (BHA) occurs, severely impacting the network's capacity to handle data. Consequently, a network security method against the BHA node must be designed. In this article, we take a look at Ad-hoc On-Demand Distance Vector (AODV), a recently revamped routing system that integrates the best features of ANN, SVM, and Artificial Bee Colony (ABC). A unique aspect of the suggested model that aids in attacker identification inside the identified path utilizing the AODV routing mechanism is the integration of SVM with ANN. With this setup, we train the model with ANN, but we use the ABC fitness function and SVM to choose our training data. In data transmission, ABC's job is to improve the path from the source node to the destination node. After ABC suggests an optimal route, it sends the node's attributes and the path to the SVM model. The ANN determines whether the node is malicious or not based on those characteristics. The suggested work improves upon the existing state in terms of latency, throughput, and Packet Delivery Ratio (PDR), according to the MATLAB simulation study. A comparison with current methods, such as Decision Tree and Random Forest, is conducted to verify the system's efficacy; the results show that combining SVM with ANN is a good move for detecting BHA attackers in MANET-based IoT networks.

DOI: 10.33545/27076571.2024.v5.i2b.101

Pages: 105-109 | Views: 892 | Downloads: 365

Download Full Article: Click Here