Research on edge-intelligent IoT collaborative perception and adaptive spatial design optimization for urban digital twins

Edge collaborative perception for urban digital twins requires timely state maintenance under constrained communication resources. Existing collaborative perception methods mainly optimize detection accuracy and communication compression, whereas urban-scale 3D reconstruction methods generally assume that observations have already been acquired or transmitted. Neither directly addresses how limited edge communication resources should be allocated according to their expected contribution to persistent twin-state maintenance. To address this gap, this paper proposes a twin-update-aware communication-allocation framework for edge-enabled urban digital twins. Candidate perception queries are selected under a fixed bandwidth budget according to their predicted twin-state variation, uncertainty reduction, spatial-feedback relevance, and communication cost. An event-driven semantic Gaussian backend then performs localized semantic and identity-preserving updates in event-affected regions, while a spatial-feedback estimator derives protocol-defined operational indicators and sensitivity signals from the updated twin state. The propagated sensitivity is explicitly used in the next time step to adjust subsequent query priorities, forming a feedback-coupled communication process rather than a post-hoc feedback module. Under a multi-dataset hierarchical proxy protocol, the proposed method achieves a favorable perception–communication trade-off and improved twin-maintenance utility. On V2X-Real, it achieves 72.4% 3D mAP and 77.6% recall at 4.4 Mb/frame, while the fixed-backend evaluation yields a TSUG of 0.312 and a DSCE of 0.103. Sequential experiments further show that dynamic sensitivity propagation reduces cumulative twin-state error relative to static or sensitivity-free variants. These results concern protocol-defined twin maintenance and feedback-aware communication utility and should not be interpreted as evidence of complete urban spatial-design optimization.