碩士 國立清華大學 電機工程學系 104 In recent years, the air pollution and the change of lifestyle enhance the probability of suffering lung disease. Because most of measuring instruments are expensive and placed in hospitals, which are operated with the assistance of the professional. Measuring respiration information by radar system were able to provide the convenience for subject and save the travel time to hospital. In this study, we added the automatic position into the FPGA. It can accelerate the measurement time for the subject. First, we record the subject's process of respiration, then extracted respiratory feature by FSLW model. We can use the feature to estimate and judge the condition of the respiratory system. Until now, pulmonary spirometer is one of the main instruments to measure respiratory indexes. We set the indexes as standard value, which are extracted by pulmonary spirometer, and set FSLW's indexes as the input conditions to classify respiratory ability and predict respiratory indexes. Where we use the support vector machine (SVM) to learn and classify respiratory ability, and use the adaptive neuro-fuzzy inference system (ANFIS) to learn and predict respiratory indexes. Through the results of the experiment, it indicated that the proposed the respiratory index extracted by radar system is reliable. We combine the gait system to increase input conditions of SVM and ANFIS. According to the results, the cooperation with gait system has improved the classification accuracy and correlation with pulmonary spirometer.