In　recent　years,　smart　home　technologies　have　started　to　be　widely　used,　bringing　a　great　deal　of　convenience　to　people＇s　daily　lives.　At　the　same　time,　privacy　issues　have　become　particularly　prominent.　Traditional　encryption　methods　can　no　longer　meet　the　needs　of　privacy　protection　in　smart　home　applications,　since　attacks　can　be　launched　even　without　the　need　for　access　to　the　cipher.　Rather,　attacks　can　be　successfully　realized　through　analyzing　the　frequency　of　radio　signals,　as　well　as　the　timestamp　series,　so　that　the　daily　activities　of　the　residents　in　the　smart　home　can　be　learnt.　Such　types　of　attacks　can　achieve　a　very　high　success　rate,　making　them　a　great　threat　to　users＇　privacy.　In　this　paper,　we　propose　an　adaptive　method　based　on　sample　data　analysis　and　supervised　learning　(SDASL),　to　hide　the　patterns　of　daily　routines　of　residents　that　would　adapt　to　dynamically　changing　network　loads.　Compared　to　some　existing　solutions,　our　proposed　method　exhibits　advantages　such　as　low　energy　consumption,　low　latency,　strong　adaptability,　and　effective　privacy　protection.　©　2017　by　the　authors.