IoT　sensors　have　already　penetrated　into　extremely　broad　fields　such　as　industrial　production,　smart　home,　environmental　protection,　medical　diagnosis,　and　bioengineering.　Although　efficient　data　fusion　helps　improve　the　quality　of　intelligent　services　provided　by　the　Internet　of　things,　because　the　perceived　data　carry　the　sensitive　information　of　the　perceived　object,　the　data　fusion　process　is　prone　to　the　risk　of　privacy　leakage.　To　this　end,　in　this　paper,　we　proposed　a　privacy-enhanced　federated　learning　data　fusion　strategy.　This　strategy　adds　Gaussian　noise　at　different　stages　of　federated　learning　to　achieve　privacy　protection　in　the　data　fusion　process.　Experimental　results　show　that　this　strategy　provides　better　privacy　protection　while　achieving　high-precision　IoT　data　fusion.　©　2022　Qingxin　Lin　et　al.