Optimizing Context Caching Using a Novel Hybrid Strategy for Dynamically Monitoring Access Probability

The Internet of Things (loT) ecosystem is rapidly evolving, offering unprecedented opportunities to address complex challenges in various sectors, including mobile services through intelligent decision-making and actuation. Central to the functionality of context-aware IoT applications in mobile environments is the ability to access and utilize context in-formation in a timely manner, a task facilitated by Context Management Platforms (CMPs) a cloud based system. This paper introduces “CAPME” (Context Access Probability Monitoring Engine), a novel hybrid strategy designed to optimize the caching of context information in CMPs. We introduce Probability of Access (PoA) as a key metric to ensure that context information critical for real-time decisions and actuations by mobile IoT applications are readily available. CAPME integrates Multi-Attribute Utility Theory (MAUT) with a Deep Q-Network (DQN) reinforcement learning algorithm, offering a dynamic system that precisely assesses the utility of various context attributes. This hybrid algorithm considers factors such as usage frequency, cost, timeliness, context freshness, and quality of context, enabling adaptive prioritization that aligns with the ever-changing de-mands of mobile IoT environments and applications. Through a comprehensive experimental evaluation utilizing data obtained from a real-world mobile IoT applications we showcase the capability of CAPME to significantly improve the performance of context caching. Results indicate marked enhancements in cache hit rates, reduced response times/latency, and lower operational costs, demonstrating CAPME's effectiveness over conventional caching methods used to support context-aware IoT applications in mobile environments.