QTC: A Dynamic Network Simulation and Low‑Interference Packet Capture Solution for Mobile Video App Testing

In the mobile Internet era, network conditions exhibit three main characteristics: diverse access networks (2G/3G/4G/Wi‑Fi), complex real‑world scenarios (streets, stadiums, cars, subways), and dynamic random changes (strong, weak, or no network).

To address video playback anomalies, iQIYI’s testing team extended the open‑source ATC (Augmented Traffic Control) with dynamic configuration capabilities, creating the QTC (QIYI Traffic Control) tool that can simulate both static and dynamic network variations while providing a low‑loss automated packet‑capture solution for high‑precision performance testing.

Background : Functional testing under abnormal network conditions is essential for ensuring product correctness and user experience. Existing tools fall into hardware (network impairers) and software (commercial tools like Fiddler/Charles, open‑source ATC, clumsy) categories. Hardware offers high precision but is costly; software is cheaper but less accurate and struggles with dynamic scenarios.

Problems :

Need a stable tool to simulate specified, continuously changing network conditions.

Need a low‑interference automated packet‑capture solution.

The ideal tool would accept quantified metrics (bandwidth, latency, loss, jitter, error rate) and continuously reproduce the corresponding network dynamics.

Solution Overview : The QTC architecture (see Fig. 2) combines hardware (dual‑NIC workstation, wireless access) and software (Linux + Python 2.7) components. The system can be deployed in under five minutes and operates in a LAN created by a QTC hotspot or switch.

Product Form :

Hardware: dual‑NIC workstation + wireless access device.

Software: Linux OS, Python 2.7.

Deployment time: < 5 minutes.

Applicable scenarios: devices that can join the QTC LAN.

Not suitable for bandwidth < 64 bps or parameter switch time < 2 s.

Usage : A browser‑based configuration UI works across Android, iOS, and Windows. Test devices configure parameters directly without additional equipment, enabling on‑the‑fly mobile testing within the wireless hotspot coverage.

Technical Details :

Dynamic scene configuration: new data structures store dynamic scenes, a dedicated process updates network settings, and inter‑process synchronization refreshes the UI.

Low‑loss packet capture: a mirrored switch forwards phone traffic to a PC for capture, separating application traffic from capture traffic and reducing interference (see Fig. 9).

Software automation: an HTTP monitoring service on the PC receives requests from the phone to start/stop capture, enabling fully automated packet collection.

Value : Deployed across iQIYI’s core network, QTC allows any device within the hotspot area to simulate arbitrary abnormal network scenarios, supporting “mobile testing” without cross‑device interference. It has been used for playback SDK, baseline app, TV app, and short‑video testing, adding 26 weak‑network test cases and uncovering ~10 bugs per version. The new packet‑capture setup also helped identify latency issues caused by DNS delays.

Future Plans :

Model real‑world scenarios more precisely and build a network‑scene database.

Leverage AI to mine network‑scene data and detect anomalous patterns.

Extend automated packet‑capture to other network‑request‑driven client functionalities.