?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Augmenting+Experimental+Data+with+Simulations+to+Improve+Activity+Classification+in+Healthcare+Monitoring&rft.creator=Tang%2C+C&rft.creator=Vishwakarma%2C+S&rft.creator=Li%2C+W&rft.creator=Adve%2C+R&rft.creator=Julier%2C+S&rft.creator=Chetty%2C+K&rft.description=Human+micro-Doppler+signatures+in+most+passive%0D%0AWiFi+radar+(PWR)+scenarios+are+captured+through+real-world%0D%0Ameasurements+using+various+hardware+platforms.+However%2C%0D%0Agathering+large+volumes+of+high+quality+and+diverse+real+radar%0D%0Adatasets+has+always+been+an+expensive+and+laborious+task.+This%0D%0Awork+presents+an+open-source+motion+capture+data-driven+simulation+tool+SimHumalator+that+is+able+to+generate+human+microDoppler+radar+data+in+PWR+scenarios.+We+qualitatively+compare%0D%0Athe+micro-Doppler+signatures+generated+through+SimHumalator%0D%0Awith+the+measured+real+signatures.+Here%2C+we+present+the+use+of%0D%0ASimHumalator+to+simulate+a+set+of+human+actions.+We+demonstrate+that+augmenting+a+measurement+database+with+simulated%0D%0Adata%2C+using+SimHumalator%2C+results+in+an+8%25+improvement+in%0D%0Aclassification+accuracy.+Our+results+suggest+that+simulation+data%0D%0Acan+be+used+to+augment+experimental+datasets+of+limited+volume%0D%0Ato+address+the+cold-start+problem+typically+encountered+in+radar%0D%0Aresearch.&rft.subject=Passive+WiFi+Sensing%2C+micro-Dopplers%2C+activity+recognition%2C+deep+learning%2C+simulator&rft.publisher=IEEE&rft.date=2021-05-17&rft.type=Proceedings+paper&rft.publisher=IEEE+Radar+Conference&rft.language=eng&rft.source=+++++In%3A++Proceedings+of+the+2021+IEEE+Radar+Conference+(RadarConf+'21).++++IEEE%3A+Atlanta%2C+GA%2C+USA.+(2021)++++(In+press).++&rft.format=text&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10120445%2F1%2FRadarConf2021_Final_Version.pdf&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10120445%2F&rft.rights=open