| Mobile crowdsensing is an emerging data collection paradigm,which uses mobile users carrying smart devices to collaborate to complete complex tasks that are difficult for individuals to handle.Compared with the traditional wireless sensor network,the mobile crowdsensing network regards users carrying smart devices as dynamic sensor nodes,which has the advantages of wide sensing range,large sensing scale and low maintenance cost.It has broad prospects and application potential in the fields of environmental detection,intelligent transportation,public safety and smart cities.As the applications of mobile crowdsensing continue to expand,the number of publishers using mobile crowdsensing networks to obtain perception data continues to increase.In order to obtain better quality perception resources,competition is intensifying,which seriously affects the fairness and stability of the perception network.However,the existing theoretical research mostly focuses on mobile crowdsensing in the context of a single publisher.The existing perception framework cannot handle the competition between publishers.To the best of our knowledge,there has been no research on the perception framework in the context of multiple publishers.In addition,there is also a lack of research on the task assignment mechanism in the context of multiple publishers.This thesis mainly studies the game relationship among publishers,the fairness of the platform and the task assignment mechanism in the context of multiple publishers.The main contributions and innovations of this thesis are as follows:(1)We comprehensively analyze the power,strategy and stability of cooperation between publishers,thereby constructing a perception framework based on cooperation and changing the state of competition between publishers to cooperation.The fairness and stability of the platform is the basis of the publisher to adopt a cooperative strategy.Therefore,we propose a benefit redistribution algorithm to maintain the fairness of the platform.The benefit redistribution algorithm adjusts the contribution realization of the publishers through the redistribution of benefits,thereby reducing the overall sense of relative exploitation of the platform and maintaining the fairness of the platform.Subsequent simulation experiments were carried out,and simulations in real scenarios showed that the cooperation-based perception framework can effectively reduce the platform's perception costs;meanwhile,the benefit redistribution algorithm can effectively reduce the overall sense of relative exploitation of the platform by 30% to60%,and effectively improve the fairness of the platform.(2)Fully analyzed the new characteristics of task distribution of the context of multiple publishers: task diversity,perceived resource scarcity,and high requirements for data quality.In response to these new features,develop a reasonable data quality evaluation model and incentive mechanism.The former effectively evaluates participants' perception ability and data quality,while the latter can motivate participants to actively participate in perception tasks.On the basis of these works,we convert the task assignment problem of the context of multiple publishers into an optimization problem,and maximize the total quality reward ratio as the optimization goal.Three task assignment strategies are designed based on different angles.They are QFA algorithm based on greedy algorithm,GGA-Q algorithm combined with greedy and genetic algorithm and DA algorithm based on local search.Simulation results show that these three algorithms have greater advantages in terms of data quality and perceived cost than other comparison algorithms.In particular,the task distribution results generated by the GGA-Q algorithm have improved performance by more than 30%compared to the comparison algorithm;the DA algorithm shows a clear advantage in reducing the perceived cost while taking into account the data quality when the participants are scarce. |
PDF Full Text Request