Exercise electrocardiography (the exercise ECG test or the stress EGG test) is one of the most popular and the most important non-invasive diagnostic tests in the field of cardiovascular disease. Arrhythmia analysis is an important part of the exercise ECG. A new approach to arrhythmia analysis is proposed in this thesis. 12 lead ECG signal is first reduced into three orthogonal channels which contain all the power of ECG. The orthogonalization process, an online Singular Value Decomposition (SVD) algorithm, maintains that these channels are free from both baseline wander and EMG noise. The third output channel has very low power with respect to first two. Making use of the orthogonality of these new channels. Total Power Signal (TPS) is calculated by summing the squares of orthogonalized channels. Employing the first two channels in TPS yields 92-99% of ECG power contained in all channels. Any arrhythmic behaviour during exercise test effects the TPS. In order to obtain the fiducial points of QRS complex first derivative of TPS is calculated. The method is compared with an algorithm proposed previously. The method and the algorithm are tested on 22 complete stress ECG test each with a duration between 9.5 to 26.5 minutes.