Multiple Regression Network Performance Evaluation Based On Information Entropy Theory For The Internet Of Things

This paper addresses network performance evaluation for the Internet of Things.Auto-ID Lab at Massachusetts Institute of Technology first proposed the concept ofthe Internet of Things in1999. The Internet of Things collects information with radiofrequency identification, sensors, global positioning systems, wireless sensornetworks, and so on. According to predetermined protocols, the Internet of Things,through the access network, builds an intelligent connection between all the thingsand the Internet, so that it can realize the monitoring, tracking, localization,identification and management intelligently and autonomously. Nowadays, theInternet of Things are widely used in smart grid, intelligent transportation, smarthome, smart logistics, smart medical, environmental monitoring and ecologicalagriculture, and so on. The wide and common use of the Internet of Things promotesthe study of network performance evaluation for the Internet of Things. Scientific andrational network performance evaluation is useful for users to understand theoperating conditions of the network timely and accurately. And the evaluation resultscan provide basis for decision making to optimize network configuration and improvenetwork performance.The Internet of Things is a heterogeneous network, which includes wirelesssensor networks, RFID, GPS, M2M and other networks at the underlying network.This brings a variety of challenges for the network performance evaluation of theInternet of Things. This paper, for the heterogeneity of the Internet of Things,proposes an application-oriented multiple regression network performance evaluationbased on information entropy theory. This evaluation model, from the perspective ofapplication and service, combines qualitative evaluation method and quantitativeevaluation method. Then, it gives the four levels of network performance, andproposes network performance evaluation value and cumulative network performanceevaluation value. The model calculates the weight of each attribute with entropymethod. According to the established model, this paper evaluates the performance ofthe environmental monitoring platform. It gives network performance levels of theInternet of Things at different times, and calculates the corresponding networkperformance evaluation value and cumulative network performance evaluation value.Experimental results show that the proposed model can provide reliable, accurate andcredible network performance evaluation for the Internet of Things. Finally, theexperimental results were analyzed and evaluated, and the consistent contrast,between qualitative evaluation results and quantitative evaluation results, shows onceagain the reliability, accuracy and credibility of the proposed network performanceevaluation model in this paper.