Automated Performance Testing Solutions for Android and iOS Apps

Mobile app performance measurement often suffers from three core issues—data accuracy, reliability, and volume—while requiring minimal manual effort and cross‑platform knowledge; these constraints drive the need for automated, configurable testing solutions.

Ordered UI automation scripts are essential for meaningful comparisons, so a simple, configuration‑driven service that abstracts complex scripts and supports offline debugging is proposed.

Because Android devices are numerous, a remote device‑rental platform (e.g., Baidu’s RMTC) is leveraged to provide scalable device pools for parallel testing.

The business requirements are summarized as: execute ordered UI automation scenarios, collect real‑time precise data, deliver remote device services with ample sampling, and support both Android and iOS with offline debugging capabilities.

Implementation goals include a lightweight configuration service to trigger performance tests, integration of performance data with page‑driven actions, email reports with aggregated graphs, automatic app installation and special device selection, and dual‑platform offline debugging.

Android solution: Two modes are offered—(1) a PC client that enables local debugging without rooting the device, and (2) remote batch testing across many devices. The PC client wraps ADB commands to gather CPU, memory, response time, power consumption, and frame‑rate metrics, while the Luban ecosystem (PC, App, Platform) handles script sequencing, UIAutomator actions, app download, and HTTP data transmission.

The remote batch testing architecture overcomes cross‑network challenges by tunneling device connections through the RMTC platform, enabling large‑scale device scheduling (e.g., up to 20 devices per 8‑core server) and load‑balanced task dispatch via the Luban server API.

Typical user workflow: (1) fill in the scenario configuration, (2) select multiple devices to run the test case, and (3) view performance trend charts for each device.

iOS solution: Utilizes Facebook’s open‑source WebDriverAgent (WDA) on a Mac to control 4‑6 iOS devices via iproxy; a binary launcher (launchclient.py) receives HTTP‑delivered configurations and drives the devices. Performance data is collected through an embedded SDK, capturing CPU, memory, FPS, traffic, and power metrics.

For competitor apps, a jail‑broken device is used to decrypt and re‑sign the IPA, allowing Instruments to capture performance data; binary files are parsed to extract readable metrics, which are then uploaded to the Luban backend for analysis.

Finally, the platform generates email reports that compare core use‑case performance across multiple competitors, presenting clear trend charts and delta analyses.

The article concludes by inviting readers to engage with the authors for further collaboration on automated mobile performance testing.