A recursive learning technique for improving information processing through message classification in IoT-cloud storage

摘要

Processing and accessing distributed information is a prominent requirement for Internet of Things (IoT) in supporting business and consumer applications to improve accessibility. As the volume of information is being stored and processed is hefty, message classification is challenging in a mobile environment. This also results in prolonged processing delays and backlogs. In order to bridge the gap between message classification and request processing, this paper proposes a classification technique that operates on the basis of request-prioritized recursive learning for ease of message identification and service mapping. This learning technique predicts the type of information and its attributes through intensive learning and services them based on priority to minimize retrieval time. The priority of message servicing relies on the error obtained during the learning process. Despite an increasing number of user requests, attributes associated with each are bundled independently to provide an instant response. Prioritization accounts for the minimum number of error states while learning a message with different attributes to curtail prolonged response time. Error in the learning process is evaluated through a numerical analysis for different learning scenarios of the classified messages. The proposed learning-based access and retrieval technique were analyzed using the metrics of request backlogs, response time, caching delay, and the rate of utilization. The results of experiments verified the effectiveness of the proposed technique in terms of minimizing response time, backlogs, and caching delays, and improving utilization.