The invention relates to a method (100) that enables the following steps to be performed in any order to ensure that at least one attack to be made to any at least one OSI layer and/or to any at least one sub-layer of said OSI layer in communication systems can be predicted/detected and thus precautions are taken in advance; - Creating the training dataset by determining the machine inputs (G) and machine outputs (Ç) in the first step - Making the training datasets understandable by the machine (computer) in the second step - Designing machine learning in the third step - Creating the validation dataset in the fourth step - Training the machine (or trained machine) in the fifth step - Obtaining test data (or sets) in the sixth step - In the seventh step, estimating (or testing) the machine outputs (Ç) using the machine trained in the fifth step and the test input set obtained in the sixth step.