To　solve　the　problems　of　incomplete　network　attack　risk　assessment　and　randomization　of　IP　address　hopping　in　dynamic　defense,　the　optimal　dynamic　defense　strategy　selection　method　based　on　network　attack　surface　evaluation　is　proposed.　The　network　attack　surface　evaluation　model　based　on　the　analytic　hierarchy　process　(AHP)　and　cloud　model　is　studied,　and　the　attack　surface　evaluation　index　and　its　quantitative　method　are　designed.　A　method　for　evaluating　the　global　importance　of　host　assets　based　on　the　K-CORE　algorithm　is　proposed,　and　different　hopping　frequencies　are　set　by　the　importance　of　host　assets.　Based　on　this,　the　IP　address　hopping　offensive　and　defensive　scenario　is　modeled,　and　an　optimal　hopping　strategy　selection　method　based　on　the　dynamic　game　is　proposed.　The　value　of　attack　surface　under　different　game　conditions　is　used　as　the　offensive　and　defensive　gains　parameter,　and　the　solution　is　solved　by　refining　the　Bayesian　Nash　equilibrium,　to　achieve　the　optimal　IP　address　hopping　strategy　selection.　Experiments　show　that　the　proposed　method　can　effectively　calculate　the　optimal　hopping　strategy　and　provide　meaningful　support　for　improving　dynamic　defense　capabilities.　©　2021　IEEE.