To　address　the　problem　of　QoS　degradation　during　the　vehicle　movement,　a　novel　service　migration　via　convex-optimization-enabled　deep　reinforcement　learning　(SeMiR)　method　is　proposed.　The　optimization　problem　is　decomposed　into　two　sub-problems　and　solved　separately.　For　the　service　migration　sub-problem,　an　improved　deep　reinforcement　learning　based　service　migration　method　is　designed　to　explore　the　optimal　migration　policy.　For　the　resource　allocation　sub-problem,　a　convex　optimization　based　resource　allocation　method　is　developed　to　derive　the　optimal　resource　allocation　for　each　MEC　server　under　the　given　migration　decisions,　thereby　improving　the　performance　of　service　migration.　Experimental　results　show　that　the　SeMiR　method　can　achieve　better　QoS　and　superior　service　migration　performance　than　benchmark　methods　under　various　scenarios.　©　2025　Acta　Simulata　Systematica　Sinica.　All　rights　reserved.