Arrhythmia detection and classification are critical for early diagnosis and management of cardiac disorders. In this study, we performed a nonlinear analysis of ECG signals representing up to 35 types of arrhythmias generated by the MPS450 Fluke simulator. These signals were recorded with the ADS1298 analog front-end (AFE). Eight measures were analysed such as variance, Higuchi Fractal Dimension (HFD), Katz Fractal Dimension (KFD), Detrended Fluctuation Analysis (DFA), Sample Entropy (SampEn), Approximate Entropy (ApEn), and Multiscale Entropy (MSE). The impact of window size was tested also. These measures allowed to differentiate between a normal rhythm and 20 out of 35 arrythmias as well as between 73.7% of pairs of different arrhythmias, highlighting their diagnostic potential. The best results were obtained for 30-s window. All fractal dimension measures (KFD, HFD and DFA) and ApEn were the most important features for arrythmias classification. This study demonstrates the efficacy of nonlinear ECG signal analysis in characterizing arrhythmias and emphasizes the importance of advanced simulation and acquisition systems like the MPS450 Fluke and ADS1298 for cardiac research and clinical applications, i.e. integration of an application based on nonlinear measures with wearable ECG devices for automated arrhythmia classification.