The heart as one of the major organs of the body plays an important role in human health. According to the World Health Organization (WHO) cardiovascular diseases (CVD) are the first cause of death in the world. This issue points the importance of further care of the heart out. All heart activities are done and controlled by the electrical conduction system of the heart through generating appropriate electric pulses. Appropriate generation of these pulses causes on time and appropriate contraction of the four major muscles of the heart, so heart can work properly. The heart condition can be monitored and checked by studying the electrical signals produced by these pulses. The electrical signal corresponding to the heartbeats is called electrocardiogram (ECG) signals. ECG can be detected by electrodes placed on the skin. Compared with manual observation, automatically processing this signal with electronic equipments not only is more accurate and faster but also makes it possible that by using it in portable medical equipment; the patient's condition can be monitored at any moment. Even some of these equipments after detecting the failure can do appropriate treatment in a timely manner. As for the importance of ECG signal processing, the proper tool for this process is also important. Due to ECG signal is a nonstationary signal, wavelet transform (WT) is the best candidate for ECG signal processing. Several basic functions are provided for WT, but yet no special function has been designed for ECG signal processing. In this thesis a new orthonormal wavelet function is designed that is appropriate for ECG signal processing. In addition, implementation of the proposed wavelet needs a very simple hardware. The VHDL codes of the proposed wavelet and that of well-known Daubechies wavelet both are developed and separately implemented on several FPGAs. For example in spartan-3 XC3S50 the number of logic cells, equivalent gates and LUTs which are used in the realization of the proposed wavelet in compared with those of Daubechies wavelet are decreased by 66%, 98% and 54%, respectively and in spartan-6 XC6SLX9 slice registers are decreased by 50% and without needing any DSP block while the maximum operation frequency is almost doubled. In all of simulations the power is approximately reduced by 50%. The VHDL codes are synthesized by logic synthesis tool, Design Vision, in a 0.18um CMOS technology. The results show that combinational cells, sequential cells, power consumption and area are reduced by 95%, 59%, 75% and 85%, respectively. The proposed wavelet demonstrates that the required hardware to implement thecorresponding WT is considerably rduced. Moreover the proposed wavelet is more accurate in compared to conventional wavelet for ECG processing. The performance of the proposed wavelet for ECG feature extraction is evaluated by MATLAB simulation and implemention on spartan-6 XC6SLX9. The accuracy of the proposed wavelet in ECG feature extraction is compared with that of daubechies as a wavelet function that is widely used in ECG signal processing. The results prove that the proposed wavelet extracts the features of a noisy ECG signal with an error less than 90%. Finally a new method is introduced for threshold calculation. This method can be used in thresholding procedure that is an important step in ECG feature extraction process. Thie proposed method has high accuracy and needs simple hardware for implementation because of low complexity in its calculation. Independent of selected wavelet for ECG signal processing, the proposed method for thresholding causes increase in the accuracy of feature extraction. MATLAB simulation indicates that accuracy of ECG feature extraction of the proposed method in comparison to Donoho method is improved by 56% for Daubechies wavelet and 92% when it is used for the proposed. Keywords: Heart, ECG signal processing, wavelet transform, thresholding, hardware simplicity, FPGA.