Increasing the Security of Wireless Networks based on Machine Learning and Evolutionary Algorithms to Detect Spoofing Attack

In the spoofing attack, the attacker sends his fake request messages to the destination server computer and pretends that the requests were sent from a source that is a node in the network with a valid and reliable IP address. In this article, an intelligent method is introduced to detect the presence of several fake users. Evolutionary processing has also been used to improve the accuracy and precision of detecting the number of attackers. The artificial neural network reveals the spoofing attack by detecting the change in the signal pattern of the IP packets received from a node in the access point. The metaheuristic algorithm of imperial competition also determines the number of attackers using clustering. By extracting the features of signal strength, signal phase, and received signal energy, this algorithm can cluster the signal pattern of users with a specific IP address. In the worst case, it correctly detects attacks with a probability of 98. As a result, if the number of clusters is more than the number of active IP addresses of the wireless network, this number of excess clusters shows the number of attackers. The simulation results of the proposed algorithm show that it has 99% precision and 98% accuracy in detecting attacks.