EpiDroid: Dependency-Guided Recomposition for Deep State Discovery in Mobile GUI Testing

2026-04-01

AI Summary

This work addresses a critical limitation in existing mobile GUI testing approaches, which often overlook semantic state dependencies among test fragments, leading to redundant exploration and poor coverage of deep application states. To overcome this, the paper proposes a black-box, plug-in testing framework that introduces, for the first time, a semantic state dependency-aware mechanism. By constructing a state dependency graph and leveraging large language models for impact inference, the framework enables deterministic replay of high-value mutable state elements. It further employs a reorganize-and-replay paradigm to iteratively refine the dependency graph, thereby transcending the constraints of conventional forward-only exploration. Integrated into both industrial-grade and state-of-the-art tools across 20 real-world applications, the approach achieves average code coverage improvements of 10–28%, yielding coverage gains equivalent to 3–4 times the baseline’s continued execution.

Abstract

The increasing scale and complexity of mobile applications make automated GUI exploration essential for software quality assurance. However, existing methods often neglect state dependencies between test fragments, which leads to redundant exploration and prevents access to deep application states. We introduce EpiDroid, a black-box, pluggable framework that augments existing explorers through semantic state dependency awareness. EpiDroid distills raw traces into stable test fragments to extract underlying dependencies. It then employs a Recomposition-Replay paradigm to perform impact reasoning via LLM and deterministic replay on high-value mutable state elements. Through iterative feedback, EpiDroid refines the state-dependency graph to systematically reach deep application states. We integrated EpiDroid into both industrial and state-of-the-art research tools and evaluated it on 20 real-world apps. The results show that EpiDroid consistently improves the performance of all baselines, increasing average code coverage by 10--28\% and delivering 3--4$\times$ more coverage gain compared to continuing the baselines alone from the same starting point. This demonstrates that dependency-guided recomposition unlocks deep states that forward exploration cannot access, irrespective of additional budget.

Problem

• mobile GUI testing
• state dependencies
• deep state discovery
• automated exploration
• test redundancy

Innovation

• state dependency
• recomposition-replay
• deep state discovery
• GUI testing
• LLM-guided reasoning