Detection of Attacks and Anomalies in the Internet of Things System using Neural Networks Based on Training with PSO Algorithms, Fuzzy PSO, Comparative PSO and Mutative PSO

Integration and diversity of IOT terminals their applicable programs make them more vulnerable to many intrusive attacks. Thus, designing an intrusion detection model that ensures the security, integrity, reliability is vital. Traditional technology has disadvantages low rates weak scalability cannot adapt complicated changing environment Internet Things. Hence, one most widely used traditional methods use neural networks also evolutionary optimization algorithms train can be efficient interesting method. Therefore, in this paper, we PSO algorithm network detect attacks abnormalities system. Although benefits, some cases it may reduce population diversity, resulting early convergence. Therefore,in order solve problem, modified with a new mutation operator, fuzzy systems comparative equations. The proposed method was tested CUP-KDD data set. simulation results article show better performance 99% accuracy detecting different malicious attacks, such as DOS, R2L, U2R, PROB.