Lanzhou Terminal 3: The Unified Airport Security System Blueprint

Architecture and Airport IT Solutions Integrations

The solution merges Axxon One (large-scale AI-driven video analytics) and Axxon PSIM (access control, intrusion management, workflow scripting, and select facial recognition use cases) into a unified command-and-control platform. Together, they form a unified airport security system optimized for real-time situational awareness and coordinated flight security and controls.

AxxonSoft integrated Bosch facial terminals specified in the tender, Honeywell perimeter and intrusion systems, Huawei IVS storage, and Noder door controllers. All subsystems are orchestrated via the PSIM engine and presented to operators through map-centric client workstations.

Technical Insight – Scalable Airport CCTV

To sync with IVS storage, the team used a hybrid method: JScript automation within PSIM routinely scans all configured cameras and auto-provisions corresponding camera objects on the IVS cluster via REST API, maintaining ID parity and ensuring uninterrupted recording to IVS disks. A custom video driver intercepts archive playback requests in the client and, instead of pulling from local or edge storage, requests direct RTSP playback from IVS. Operators see standard archives, while all backend heavy lifting is offloaded to the storage layer —supporting airport IT solutions scalability.

AI Analytics Supporting Flight Security and Controls

AI workloads run on two specialized clusters:

• Axxon PSIM facial recognition cluster: 6 servers — five GPU workers generating face vectors and snapshots, plus a dedicated face-search node — capable of searching seven days of history across 500+ facial channels in a terminal handling around 15M passengers annually — for strengthening flight security and controls.
• Axxon One analytics cluster: 8 servers, each with three NVIDIA Tesla GPUs, processing 50 cameras per node and running neural-tracker analytics for appearance-based similarity search, intrusion detection, directional people counting, crowd-density estimation, and other AI-driven airport security system functions.

Initial similarity search queries could take up to 2 minutes to produce first hits across high-volume datasets. By optimizing GUI-level query orchestration and parallelizing jobs across the 8 GPU nodes, engineers cut time-to-first-result to 30 seconds, enabling operators to begin reviewing hits while additional matches populate in background streams.

Custom Web Interface for Airport Security Operations

A key tender requirement was a web-based unified portal, merging data from Axxon One and PSIM into a single browser UI with integrated maps and dashboards — removing the need for managers to rely on desktop clients. To deliver this, AxxonSoft partnered with local R&D resources, whose 10-person “MVP squad” (five front-end, five back-end) produced the first functional prototype in roughly two weeks using AxxonSoft APIs and sample/mock datasets.

The web portal aggregates facial recognition events, face-search results, access-control and intrusion alarms, AI events (crowd density, line-crossing, neural-tracker detections), and occupancy analytics into configurable dashboards and map-based widgets. Development began with a fake-data MVP to validate UI logic and analytics flow with the airport. Once approved, the platform was connected to live Axxon PSIM and Axxon One servers, turning it into a real-time operational dashboard that can be cloned for other secure airport deployments simply by updating the map and branding.

Implementation and Results

Deployment followed a staged rollout aligned with aviation authority acceptance tests — starting with a 200-camera pilot and a small group of controllers, then scaling to 4,000 cameras and 200 controllers, then 6,000 cameras and 300+ controllers, and finally 12,000+ cameras and 450+ controllers at formal project completion.

Commercial and Technological Advantages

By selecting AxxonSoft, Lanzhou Airport consolidated airport CCTV, access control, intrusion detection, and advanced AI analytics into a single unified airport security system, replacing the risk and complexity of multi-vendor, multi-system integration. This simplifies procurement, support, and lifecycle management while providing a single vendor accountable for all mission-critical security and analytics functions in Terminal 3.

The platform is open-architecture, supporting deep integrations with controllers, facial terminals, intrusion panels, GIS platforms, and clustered storage systems via open APIs and SDKs. Coupled with powerful engines for facial recognition, appearance-based similarity search, crowd analytics, and behavior detection, the solution is both future-proof and aligned with emerging aviation compliance standards.

Reusable Airport Solutions Blueprint for Future Airports

Beyond meeting Lanzhou’s operational requirements, this deployment now serves as a repeatable blueprint for future airport builds. The custom web portal can be rapidly adapted to new sites by connecting it to AxxonSoft APIs and swapping in new maps and branding, while all underlying integrations with CCTV, access control, intrusion, and AI analytics remain unchanged.

This integrated airport security system — combining PSIM, VMS, AI analytics, and a unified web front end — can be replicated with minimal lead time at other major hubs. The LHW Terminal 3 deployment stands not only as a flagship reference but also as a scalable, data-driven model for next-generation airport security and operations worldwide.