In　recent　years,　network　virtualization　has　been　introduced　to　wireless　networks　due　to　its　great　success　in　wired　networks　as　well　as　the　rapid　developing　of　wireless　networks.　Resource　management　is　always　a　key　issue　in　virtualized　wireless　networks.　In　this　paper,　under　the　resource　auction　architecture　for　resource　management,　we　focus　on　the　optimal　bidding　actions　of　the　service　providers　(SPs).　With　the　objective　of　maximizing　the　SP＇s　total　reward,　which　is　composed　of　the　user　service　reward　and　the　resource　utilizing　costs,　the　optimal　bidding　behavior　decision　scheme　with　consideration　of　caching　and　spectrum　utilization　costs,　which　is　called　B-CaB,　is　proposed　in　this　paper.　The　Markov　decision　process　with　unknown　parameter　(MDP-UP)　model　is　adopted　in　our　formulation,　and　then　we　reformulation　it　as　a　partially　observable　Markov　decision　process,　to　deal　with　the　challenging　problem　that　the　state　transition　parameter　is　not　always　known　or　observable.　Extensive　simulation　results　are　presented　to　demonstrate　the　significant　performance　improvement　of　the　proposed　scheme.