The identification of cables is becoming increasingly important due to the higher complexity of technical equipment and installations. In this letter, we propose a method for identifying coaxial communication cables based on the scattering parameters at 101 frequencies for 16 coaxial communication cables with a characteristic impedance of 50 $\Omega$ and various lengths, dimensions, and connector types. For classification, we consider the ready cable, with its specific length and type as a whole, as a unique class. The support vector machine (SVM) algorithm is used because of its ability to process high-dimensional spaces with linearly inseparable data. The achieved classification results when applying the principal component analysis on a database, including the input port reflection's magnitude values, show a testing accuracy of 100% with only five principal components. The method is suitable for realization as an embedded system.