Response Time Prediction of IoT Service Based on Time Similarity

Abstract

In the field of Internet of Things (IoT), smarter embedded devices offer functions via web services. The Quality-of-Service (QoS) prediction is a key measure that guarantees successful IoT service applications. In this study, a collaborative filtering method is presented for predicting response time of IoT service due to time-awareness characteristics of IoT. First, a calculation method of service response time similarity between different users is proposed. Then, to improve prediction accuracy, initial similarity values are adjusted and similar neighbors are selected by a similarity threshold. Finally, via a densified user-item matrix, service response time is predicted by collaborative filtering for current active users. The presented method is validated by experiments on a real web service QoS dataset. Experimental results indicate that better prediction accuracy can be achieved with the presented method.

Novel Resource Allocation Algorithms for the Social Internet of Things Based Fog Computing Paradigm

“…Accordingly, the IoT technologies make it possible to provide new services to end-users while establishing social relationships in an autonomous way. Similar to what happens with social networks among humans, smart objects in the IoT connected each other asymmetrically with their preference similarity and interests [1][2][3].…”