We　designed　a　system　perception　model　under　edge　computing　that　supports　Local　Differential　Privacy(LDP).　To　address　the　privacy　protection　challenges　during　the　data　submission　phase　in　Mobile　Crowdsensing(MCS)　and　the　increased　costs　associated　with　privacy　protection,　we　introduced　an　edge　computing　architecture　into　the　MCS　framework.　Based　on　local　differential　privacy,　we　developed　the　MCS-MADP　algorithm,　which　considers　the　attribute　relationships　of　user-submitted　data.　Experimental　results　show　that,　under　the　same　privacy　budget,　the　MCS-MADP　algorithm　improves　accuracy　by　an　average　of　67.75%　compared　to　the　LoPub　algorithm　and　by　an　average　of　139.89%　compared　to　the　PrivKV　algorithm.　Additionally,　the　MCS-MADP　algorithm　enhances　data　accuracy　while　reducing　the　required　data　volume.　©　2024　Copyright　held　by　the　owner/author(s).　Publication　rights　licensed　to　ACM.