An Improved Troika Framework for Heart Rate*Monitoring Using Wrist- Type Ppg Signals

A hear rate monitoring system using wrist-type PPG signals was developed. The developed system was built upon the TROIKA framework~[1]with several improvements from both effectiveness and efficiency. One of the key steps of the TROIKA framework is the sparse signal reconstruction( SSR),which aims at obtaining high-resolution spectrum of the PPG signals. As observed by our experiments,the performance of the TROIKA framework is sensitive to the presence of motion artifacts( MA) in the stage of spectrum peak tracking. To overcome this problem,the spectrum outputted was further corrected by SSR by removing MA using the acceleration data. In addition,several strategies were proposed including initialization correction,tracing back and exploiting priori knowledge in the verification step,to enhance the performance of the spectrum peak tracking. Experimental results show that our developed system gives 2. 24 errors on average on 12 training samples,which is slightly better than the TROIKA method( 2. 34 reported in~[1]).