Vehicles　in　vehicular　ad-hoc　networks　(VANETs)　are　required　to　continuously　broadcast　sensitive　data　including　coordinates　and　speeds.　Such　practices　make　sensitive　data　susceptible　to　eavesdroppers,　undermining　location　privacy.　Conventional　strategies　attempt　to　preserve　location　privacy　through　different　privacy　enhancing　techniques　such　as　encryption,　silent　periods,　and　chaff　messages.　However,　existing　approaches　fail　to　simultaneously　ensure　driving　safety　and　location　privacy,　particularly　against　semantic　linking　attacks　by　machine-learning-enabled　adversaries.　To　this　end,　this　paper　proposes　the　trajectory　converged　chaff-based　mix　zone　(TCCM)　strategy.　It　generates　chaff　messages　that　imitate　real　vehicle　trajectories　to　confuse　eavesdroppers　while　maintaining　low　communication　overhead,　thereby　enhancing　location　privacy　without　compromising　vehicle　safety.　Additionally,　the　TCCM　strategy　incorporates　a　genetic　algorithm　to　optimize　mix　zone　placement　and　ensure　a　balance　between　privacy　protection　and　resource　efficiency.　We　implemented　a　VANET　simulation　and　two　adversary　attack　algorithms　to　evaluate　the　performance　of　our　scheme.　Reportedly,　the　TCCM　strategy　reduces　trajectory　traceability　by　at　least　24.6%　compared　with　conventional　mix　zone　strategies　while　maintaining　vehicle　safety.　Additionally,　the　chaff　messages　of　the　TCCM　strategy　incur　54%　less　communication　overhead　than　existing　chaff-based　schemes.　　©　2013　IEEE.