To　solve　the　problem　that　the　number　of　signal　sources　is　more　than　the　number　of　sensors　when　Gaussian　colored　noise　exists,　a　fast　wideband　DOA　estimation　algorithm　based　on　improved　fourth-order　cumulant　is　proposed.　The　complexity　of　the　traditional　fourth-order　cumulant　algorithm　is　high　which　influences　the　real-time　demand.　This　paper　proposes　a　fast　algorithm　to　improve　the　traditional　methods　by　eliminating　the　redundant　information　of　the　fourth-order　cumulant,　and　it　can　reduce　the　computational　complexity　and　the　running　time　effectively.　And　the　algorithm　using　the　signal　subspace　method　doesn＇t　need　angle　pre-estimation　when　constructing　the　focusing　matrix,　so　it　can　improve　the　estimation　accuracy　obviously.　The　simulation　results　show　the　effectiveness　and　robustness　of　the　algorithm.