Biometric　authentication　is　pivotal　in　protecting　user　privacy　and　ensuring　smartphone　security.　Recent　research　indicates　that　the　vibration　response　of　hands　can　serve　as　a　biometric　for　user　authentication　on　mobile　devices.　However,　the　various　grip　gestures　of　users　can　introduce　significant　noise,　disrupting　hand　biometrics-related　signals　and　compromising　authentication　performance.　This　paper　introduces　HandID,　an　unobtrusive　and　gesture-independent　user　authentication　method　for　smartphones　that　does　not　require　large　amounts　of　data　from　different　gestures.　Unlike　existing　methods,　HandID　does　not　require　users　to　maintain　specific　gestures　or　interact　with　the　touchscreen.　It　utilizes　the　built-in　vibration　motor　to　generate　active　vibrations　and　sense　the　user＇s　hand　through　responses　captured　by　the　onboard　accelerometer.　HandID　employs　an　adversarial　learning　model　to　handle　gesture　variations　and　proposes　a　novel　hand　biometric　generation　model　to　reduce　the　enrollment　data　required　from　users.　Comprehensive　experiments　with　50　subjects　show　that　HandID　achieves　an　authentication　accuracy　of　92.5%,　with　a　false　acceptance　rate　(FAR)　of　5.2%　and　a　false　rejection　rate　(FRR)　of　5.6%.　Security　analyses　demonstrate　that　HandID　is　resistant　to　zero-effort,　s　tatistical,　a　nd　h　ill-climbing　a　ttacks,　a　nd　a　u　sability　study　indicates　high　user　acceptance　of　HandID.　©　2025　ACM.