This　letter　investigates　the　problem　for　Spectrum　Sensing　Data　Falsification　(SSDF)　attacks　in　the　Cognitive　Internet　of　Vehicles　(CIoV)　network.　The　high-speed　movement　of　Vehicle　Users　(VUs)　leads　to　rapid　changes　in　Channel　State　Information　(CSI)　and　location.　This　leads　to　unstable　detection　probabilities　and　unstable　probabilities　of　reporting　errors.　These　unstable　probabilities　increase　the　error　rate　of　traditional　methods　to　identify　VUs　as　Malicious　Vehicle　Users　(MVUs).　And　high-speed　movement　makes　it　difficult　to　detect　MVUs,　which　may　result　in　massive　MVUs＇　attacks.　To　address　the　above　problems,　this　letter　establishes　the　Cooperative　Spectrum　Sensing　(CSS)　and　spectrum　access　process　under　a　Directed　Acyclic　Graph　(DAG)　blockchain　framework,　models　MVUs＇　attack　strategy　selection　which　is　determined　by　revenue　as　an　evolutionary　game,　and　proposes　a　smart　contract　that　changes　the　mining　difficulty　of　VUs　based　on　the　correctness　of　local　spectrum　sensing　decisions　to　influence　VUs＇　revenue.　Finally,　the　simulation　results　verify　the　theoretical　analysis　and　prove　that　the　proposed　method　is　superior　to　the　traditional　method.