云边协同收费系统多级容灾与状态同步机制研究

Research on Multi-Level Disaster Recovery and State Synchronization Mechanism for Cloud-Edge Collaborative Toll Collection Systems

摘要: 针对高速公路收费系统云边协同部署后链路抖动、服务不可用、节点异常和数据不同步等问题，提出一种多级容灾与状态同步机制。首先，从链路层、业务层、计算层和数据层划分系统故障模式，构建分层容灾控制模型，并建立链路保护、异常判定和业务恢复之间的时间尺度约束关系；其次，基于WebSocket双连接心跳探测机制，定义服务可用性向量和系统工作模式变量，构建业务层降级有限状态机，实现云站正常、疑似异常、降级判定、广场降级、恢复探测和人工确认回切等状态流转；最后，结合实站测试数据对模型进行验证。结果表明，主备切换动作最大触发时间为3 s，业务中断时间最大值为40 s，满足业务层异常判定与降级切换及恢复约束。该机制能够支撑云边协同收费系统在异常场景下实现快速降级、业务恢复和稳态回切。

Abstract: To address the problems of link jitter, service unavailability, node anomalies, and data inconsistency arising from the cloud-edge collaborative deployment of highway toll collection systems, a multi-level disaster recovery and state synchronization mechanism is proposed. First, system failure modes are classified into the link, business, computing, and data layers, based on which a hierarchical disaster recovery control model is established, together with the time-scale constraints among link protection, anomaly determination, and business recovery. Second, based on a WebSocket dual-connection heartbeat detection mechanism, a service availability vector and system operation mode variables are defined, and a finite-state machine for business-layer degradation is constructed to realize state transitions among normal cloud-station operation, suspected anomaly, degradation determination, plaza-level degradation, recovery detection, and manual confirmation for switchback. Finally, the model is validated using data collected from actual toll station tests. The results show that the maximum trigger time for the primary-backup switching action is 3 s, while the maximum business interruption time is 40 s, satisfying the constraints for anomaly determination and degradation switching and recovery at the business layer. The proposed mechanism supports rapid degradation, business recovery, and stable switchback of cloud-edge collaborative toll collection systems under abnormal conditions.