Indoor　positioning　technology　has　developed　steadily.　With　the　popularization　of　machine　learning　algorithms,　numerous　indoor　positioning　algorithms　have　emerged.　However,　most　of　the　proposed　indoor　location　algorithms　are　in　the　theoretical　and　laboratory　stages.　There　are　too　few　algorithms　for　specific　scenes.　Although　the　existing　deployment　has　been　applied　in　practice,　its　high　positioning　error　still　exists.　This　paper　proposes　a　WiFi　fingerprint　location　model　based　on　WKNN　(Weighted　K-nearest　Neighbor)　algorithm.　An　improved　WKNN　location　algorithm　based　on　local　sensitive　hash　algorithm　is　established,　which　solves　the　problem　of　poor　real-time　online　location　caused　by　the　huge　fingerprint　database.　The　algorithm　is　divided　into　two　stages:　off-line　indexing　and　on-line　positioning.　In　the　off-line　indexing　stage,　the　signature　matrix　with　reduced　dimension　is　generated　by　local　sensitive　hash　algorithm,　and　in　the　on-line　positioning　stage,　the　location　of　undetermined　sites　is　calculated　by　WKNN　algorithm.　The　results　show　that　the　RMSE　of　the　proposed　method　is　smaller　at　different　reference　distances.　Compared　with　WKNN　and　KNN,　the　method　proposed　in　this　paper　has　less　accuracy　no　matter　how　the　value　of　k　changes.　Compared　with　traditional　methods　such　as　WKNN　and　KNN,　this　method　has　better　positioning　accuracy　and　stability,　and　can　better　meet　the　current　positioning　accuracy　requirements.　©　2023　IEEE.