Evaluation Of QoS In Mobile Internet Based On Crowdsensing

Mobile Internet is the integration of traditional mobile communication network and the Internet. It provides users with more flexible network services, whenever and wherever they want. With the expansion of the mobile Internet market size, the fierce competition makes operators aware that improving QoS (Quality of Service) of mobile Internet becomes the key to survive and profit. Therefore, QoS of the mobile Internet has drawn more and more attention. The accurate evaluation of QoS of the mobile Internet has become an urgent problem.At the same time, with the explosion of mobile devices and mature model of crowdsourcing, their combination produces crowdsensing. The advantage of crowdsensing is that it can make full use of crowd power and accomplish large-scale,high-quality data collection tasks without restriction of time and space. Thus,crowdsensing has been widely used in all aspects of our daily life.This thesis focuses on the evaluation of QoS in mobile Internet based on crowdsensing.Firstly, a QoS evaluation method of mobile Internet based on user-weighted iterative algorithm is proposed. Considering the differences among users caused by the individual factors and different evaluation scenarios, users and various types of KPI are separately weighted to improve evaluation accuracy in this method. We validate its accuracy by simulation and measured data in LTE networks. The simulation results show that average evaluation result error distance of user-weighted iterative algorithm is 44.9% smaller than majority voting algorithms. The average user weight error distance is 16.7% smaller than majority voting algorithms. Based on the measured data of the mobile Internet in LTE networks, we validate and compare the accuracy between algorithms in different evaluation scenarios. The error rate of user-weighted iterative algorithm is 30% less than the majority voting algorithms in the highway and residential area.Secondly, an incentive mechanism for crowdsensing platform based on pricing strategy is proposed. The interaction between sensing task publisher and crowdsensing platforms is modeled by Stackelberg game. An iterative learning algorithm is proposed to obtain an optimal pricing strategy of the crowdsensing platforms on Nash equilibrium. Through extensive simulations, we evaluate the performance of our incentive mechanisms.In summary, this thesis provides technical support for the evaluation of QoS in mobile Internet based on crowdsensing through theoretical analysis and experiments.