Cognitive　WLAN　over　Fiber　(CWLANoF)　is　proposed　as　the　improved　architecture　of　traditional　WLANs.　In　this　paper,　the　channel　allocation　problem　for　hot　spot　in　CWLANoF　is　firstly　established　as　a　restless　bandit　problem,　which　is　solved　by　the　primal-dual　index　heuristic　algorithm　based　on　the　first　order　relaxation　due　to　less　complexity.　Moreover,　the　proposed　scheme　is　divided　into　offline　computation　and　online　allocation　so　as　to　decrease　the　complexity　farther.　Finally,　extensive　simulation　results　illustrate　the　significant　performance　improvement　of　the　proposed　scheme　compared　to　the　existing　one.