Adaptive Software Testing Framework for Internet of Things Applications

Abstract

Internet of Things applications operate across heterogeneous devices, networks, gateways, and cloud services, making conventional fixed testing insufficient for changing runtime conditions. This study proposes an Adaptive IoT Software Testing Framework, called AISTF, to improve defect detection and reduce redundant execution. The framework models device states, communication behaviour, service dependencies, sensor boundaries, and recovery events, then calculates a Dynamic Test Relevance Score using failure frequency, state novelty, dependency importance, environmental variation, execution age, and fault impact. Tests are adaptively selected, prioritised, and combined with controlled communication and sensor disturbances. The framework was evaluated over five testing cycles against fixed-suite testing, random selection, and failure-history prioritisation. AISTF achieved a cumulative defect detection rate of 96%, compared with 79%, 73%, and 86% for the comparison methods, while improving recovery-path coverage and reducing repeated tests. These findings show that context-aware adaptation can strengthen testing efficiency, resilience assessment, and quality assurance for IoT applications.