Location-Based　Service　has　moved　into　indoor　scene　by　mature　Wi-Fi　network　infrastructure　and　intelligent　handheld　terminal　device　development　in　decade　years.　Traditional　indoor　localization　systems　is　designed　by　RSSI　fingerprint　mechanism　to　estimate　user　location.　This　paper　indicated　the　limitation　of　pure　RSSI　by　analysing　the　RSSI　performance　in　experiment　and　proposed　an　RSSI-Pro　transformation　algorithm　for　radio　map　construction.　To　verify　the　desired　accuracy　an　actual　indoor　localization　system　called　LocNeedle　is　designed　to　implement　the　method　of　this　article.　LocNeedle　adopts　Bhattacharyya　distance　as　similarity　measurement　metric　to　find　the　target　user　location,　which　can　achieve　that　the　95%　of　localization　error　distance　is　under　2.5　meters　while　70%　is　under　1　meter.