Fog　computing　is　a　technology　that　can　expands　the　network　computing　mode　of　cloud　computing　and　extends　network　computing　from　the　network　center　to　the　network　edge.　It　adds　fog　layer　between　cloud　data　center　layer　and　Internet　of　Things　(IoT)　device　layer,　and　provides　data　storage,　processing,　forwarding　and　other　functions　for　devices　using　the　network　edge.　In　mobile　fog　computing　(MFC)　networks,　fog　nodes　communicate　with　end　users　through　wireless　networks.　Malicious　users　can　choose　different　attack　modes　to　attack　legitimate　users.　There　is　a　lack　of　research　on　the　subjective　choice　of　attack　modes　for　malicious　users　in　current　work.　To　solve　this　problem,　an　intelligent　attack　defense　scheme　based　on　Double　Q-learning　(DQL)　algorithm　in　MFC　is　proposed.　Firstly,　the　security　model　involving　malicious　users　in　MFC　is　described.　Based　on　Prospect　Theory　(PT),　a　static　method　of　subjective　zero-sum　game　between　malicious　users　and　legitimate　users　is　constructed.　Secondly,　a　dynamic　subjective　game　scheme　based　on　DQL　algorithm　is　proposed　to　resist　intelligent　attacks.　The　simulation　results　show　that　compared　with　the　Q-learning-based　method　for　resisting　intelligent　attacks,　the　proposed　method　can　enhance　the　security　of　MFC　network　and　enhance　the　protection　performance.　(C)　2019　Elsevier　Inc.　All　rights　reserved.