With　the　development　of　digitalization　and　informatization,　the　demand　for　location-based　services　has　shown　a　significant　growth　trend.　Indoor　is　the　occasion　where　people　stay　the　most,　indoor　localization　based　on　WiFi　RSSI　is　one　of　the　most　popular　and　efficient　localization　method.　This　paper　proposed　a　WiFi　indoor　localization　system,　which　combines　convolutional　neural　network　and　wavelet　transform.　Specifically,　we　add　wavelet　transform　to　extract　temporal　features,　and　use　one-dimensional　convolutional　neural　network　training　fingerprint　data　to　obtain　localization　result.　In　order　to　verify　the　effectiveness　of　the　designed　model,　we　set　experiments　on　real　fingerprint　dataset.　The　experimental　results　show　that　the　purpose　system　has　great　localization　accuracy　and　stability.　©　2023　IEEE.