Abstract: To improve accuracy and convenience of sleep-staging system, a structure with desktop ECG harvest center and wireless ECG paste which integrates special ASIC is presented in this paper. The sleep-staging algorithm is based on six features which are derived from RRI and HRV of single channel ECG, and these features are fed to HMM as a classifier. In which, EEMD algorithm is used to standardize RRI sequence. To improve battery life of the ECG paste, low power design method on system level and circuit level are used. The test results show that the classification accuracy of sleep-staging system is 93.2% on an average. The ECG chip occupies only 1.6 mm² on SMIC 0.18 μm process and average power consumption is only 1.3 mW, which means 70 h continuously work under a 100 mAh battery for the ECG pastes.