Recently,　effective　allocation　of　VANET　resources　is　a　key　factor　in　promoting　the　development　of　VANETs.　Due　to　high　bandwidth　costs,　poor　time　efficiency,　and　a　high　risk　of　privacy　leakage,　the　use　of　traditional　centralized　data　centers　to　analyze　massive　data　has　proven　to　be　a　difficult　task.　These　challenges　have　prompted　a　revolutionary　change　in　VANET　architectures　to　scatter　computations　from　a　centralized　data　center　to　distributed　network　edges.　Distributed　VANET　configurations　leverage　the　computing　power　of　network　edges　by　using　a　large　number　of　mobile　devices　which　frequently　exchange　data　with　the　edge　of　the　network　or　among　themselves.　However,　the　heterogeneity　and　distrust　of　the　distributed　edge　hinder　the　efficient,　reliable,　and　secure　allocation　of　VANET　resources.　In　this　paper,　we　express　the　allocation　strategy　for　both　computing　and　network　resources　as　a　joint　optimization　problem.　We　use　a　local　deep　reinforcement　learning　with　a　prioritized　experience　replay　mechanism　on　edge　nodes　and　use　the　blockchain　for　sharing　the　optimal　learning　results　to　optimize　the　overall　resource　allocation　problem.　Simulation　results　show　that　our　proposed　scheme　is　superior　to　a　current　machine　learning　approach.　©　2023,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Switzerland　AG.