In the last 4 decades, surface electromyography (sEMG) signal processing has been applied to detect local muscle fatigue, this non-invasive approach is suitable for detecting EMG signals generated by athletes in motion. Also, EMG could directly reveal the muscle’s performance like endurance and recruitment of motor units, which is hard to be obtained by other methods. With the sEMG system, we can research whether EMG signals can be used to measure muscle fatigue and how this relates to injury risk. This thesis aims to build a sensor node for sEMG to detect local muscle fatigue. An sEMG system is built for this purpose, and a physiological experiment is designed to collect sEMG signals from human muscle (Vastus Medialis) using the sEMG system. Both isometric and isotonic exercises are studied. The data analyzing method is calculating mean power spectrum frequency (MNF), median power spectrum frequency (MDF), and muscle fiber propagation velocity (MFPV) of the collected sEMG signals, because local muscle fatigue is related to MNF/MDF decrease and MFPV decrease. 5 groups of isometric exercise, wall-sit and 2 groups of isotonic exercise, cycling, are recorded. All the athletes are healthy males, around 25. The data analyzing result shows that MNF/MDF decrease is related to muscle fatigue, and MFPV changes similarly with MNF/MDF.