The　core　missions　of　IoT　are　to　sense　data,　transmit　data　and　give　feedback　to　the　real　world　based　on　the　calculation　of　the　sensed　data.　The　trust　of　sensing　source　data　and　transmission　network　is　extremely　important　to　IoT　security.　5G-IoT　with　its　low　latency,　wide　connectivity　and　high-speed　transmission　extends　the　business　scenarios　of　IoT,　yet　it　also　brings　new　challenges　to　trust　proof　solutions　of　IoT.　Currently,　there　is　a　lack　of　efficient　and　reliable　trust　proof　solutions　for　massive　dynamically　connected　nodes,　while　the　existing　solutions　have　high　computational　complexity　and　can＇t　adapt　to　time-sensitive　services　in　5G-IoT　scenarios.　In　order　to　solve　the　above　problems,　this　paper　proposes　an　adaptive　multi-dimensional　trust　proof　solution.　Firstly,　the　static　and　dynamic　attributes　of　sensing　nodes　are　metricized,　and　the　historical　interaction　as　well　as　the　recommendation　information　are　combined　with　the　comprehensive　metric　of　sensing　nodes,　and　a　multi-dimensional　fine-grained　trusted　metric　model　is　established　in　this　paper.　Then,　based　on　the　comprehensive　metrics,　the　sensing　nodes　are　logically　grouped　and　assigned　with　service　levels　to　achieve　the　screening　and　isolation　of　malicious　nodes.　At　the　same　time,　the　proposed　solution　reduces　the　energy　consumption　of　the　metric　process　and　optimizes　the　impact　of　real-time　metrics　on　the　interaction　latency.　Simulation　experiments　show　that　the　solution　can　accurately　and　efficiently　identify　malicious　nodes　and　effectively　guarantee　the　safe　and　trustworthy　operation　of　5G-IoT　nodes,　while　having　a　small　impact　on　the　latency　of　the　5G　network.　©　2023