Detecting Malicious Behaviour for SANET Based on Artificial Intelligence Algorithms

Abstract

Attack detection is important for wireless networks and communications generally. Ship ad hoc networks (SANET) are a subset of wireless networks that are vulnerable to denial-of-service attacks. These attacks are one of the main challenges facing maritime networks, specially dedicated networks because of their weak infrastructure, which makes it easier for these networks to be exposed to this type of attack. To maintain a secure connection and increase the durability of that connection, an accurate attack detection system must be built. In this paper, we used machine learning algorithms to classify data as either attack or safe. we generated the dataset by building a scenario for the SANET in the network simulator (ns-2). Ad hoc On-demand Distance Vector (AODV) was used as the routing protocol in this simulation, AODV reduces the burden on the network compared with the other protocols (reduces messages flooding in the network). Three machine learning algorithms which are Logistic Regression, Stochastic Gradient Descent and K-Nearest Neighbors were applied to the dataset and compared with each other in terms of precision in detection, the results show that the machine learning algorithms have the ability to detect attacks with higher performance. The experimental results showed that the data set that was generated with Knn as the base classifier produced the best performance in terms of classification precision by 89%