Aiming at the practical dilemma that it is difficult to collaboratively optimize fairness and service efficiency in traditional multi-server queuing scenarios, this paper proposes a hierarchical new branched queuing architecture that balances the two core objectives through grouping and constraining service order. Based on queuing theory and the permutation probability method, this paper takes the typical venue security inspection scenario as the research background, constructs a quantitative indicator system of fairness and efficiency, and compares the functional differences among single-queue, parallel multi-queue, and branched queuing modes. The measurement results show that the branched queuing architecture can reduce the expected unfairness degree by about 67% while maintaining more than 95% of the service efficiency of parallel multi-queues, which effectively solves the efficiency-fairness trade-off dilemma of traditional modes. The research conclusions can provide theoretical reference and practical schemes for the design of queuing systems in public services, transportation hubs, and other multi-server scenarios.