Microservice architecture based self-service drone aerial photography system for smart cultural tourism: Design and...

Although the low-altitude economy has been designated a strategic emerging industry, consumer-grade drone aerial photography in cultural tourism remains constrained by three coupled bottlenecks: dependence on licensed pilots, fragmented and manually orchestrated service workflows, and post-production cycles measured in hours to days. These bottlenecks jointly prevent drone aerial photography from being delivered as a standardized, on-demand consumer service. This study aims to design, implement, and empirically evaluate a software architecture that transforms professional drone aerial photography into an automated, scan-to-use consumer service, and to test whether a microservice-based design can simultaneously meet the latency, reliability, and cost requirements of mass-market smart tourism deployment. We propose KanYiKan, a distributed microservice system in which (i) an MQTT-based bidirectional channel connects the cloud scheduler with heterogeneous ground control stations, (ii) order lifecycles are governed by a deterministic finite-state machine with optimistic-locking transitions, (iii) drone assignment is formulated as a weighted multi-objective scoring problem balancing battery margin, spatial proximity, and historical reliability, and (iv) an automated video pipeline performs no-reference rule-based quality filtering, template-based semantic matching using a pre-trained vision encoder, and rendering. The system is evaluated through functional verification, JMeter-based concurrent stress testing (100-1000 users), fault-injection robustness testing, and a comparative analysis against the traditional pilot-mediated workflow. Under simulated high-concurrency loads, P99 latency of core APIs stayed at or below 312 ms and error rates below 0.05%; per-service cost dropped by over 90% (from 500-3000 CNY to 50-100 CNY), end-to-end delivery shrank from 24-72 hours to 15-30 minutes, and the pilot-per-order requirement was eliminated. All five injected fault scenarios were handled without service-wide outage. The results provide empirical evidence that a cloud-native microservice architecture, combined with FSM-driven workflow control and lightweight multi-objective scheduling, is a viable software paradigm for democratizing UAV services and offers a replicable reference design for scalable low-altitude applications in smart cultural tourism.