Research Of Fog Computing Testbed Based On Mobile Crowdsourcing

Fog computing enables users to analyze and process data nearly by extending services provided by cloud centers to edges of the Internet,which reduces the response delay of services,relieves the flow on the network backbone,remedies the lack of cloud computing,and provides foundations for various real-time applications and services under the interconnection of everything.Any distributed application based on network must be tested repeatedly for a long time before it is put into production environments,and fog computing is no exception.However,there is still lack of professional fog computing testing platforms.Most researchers usually utilize fog computing simulators or emulators to test their prototypes and there exists some problems on the authenticity and accuracy of results,so it is very important for the development of fog computing to design and construct a fog computing testing platform with fidelity.The study first investigates current research statuses of the fog computing testing field,analyzes challenges of building a fog computing testing platform,then proposes and designs the CrowdFogBed,a fog computing testing platform based on mobile crowdsourcing resources,and finally makes in-depth researches and discussions on the crowdsourcing task allocations.The main contents are as follows:(1)Fog computing testbed based on mobile crowdsourcingThe study solves the problems of diversity,heterogeneity and mobility of user layer devices in the fog computing network by introducing abundant mobile crowdsourcing devices into the fog computing testing platform,and provides real network scenarios for experimental users so as to ensure the authenticity and accuracy of experimental results and reduce risks of deploying prototypes to production environments(2)Fog experiment task allocation mechanism based on maximum task matchingTask allocation is one of the key problems needed to be resolved in mobile crowdsourcing system.In this study,we not only consider the actual needs of experimental users in CrowdFogBed,but also the preferences for task of mobile users,and then propose a fog experiment task allocation algorithm with maximum task matching based on the mechanism of double auction.Most of the existing task allocation mechanisms aim at maximizing social benefits or mobile users' benefits,and they are different from the objectives and needs of this study.Therefore,in the evaluation part,we compare the proposed algorithm with greedy algorithm,and the results show that the algorithm has a great improvement in task matching number than greedy algorithm,especially when there are fewer mobile users,the algorithm can distribute tasks more evenly to multiple users,and meet the requirements of honesty,individual rationality and budget balance(3)Fog experiment task allocation mechanism based on stable matchingThe task allocation mechanisms based on auction theories can ensure the authenticity of transactions,but cannot guarantee the stability.Based on the bi-directional auction mechanism and stable matching theory and considering the task budget constraints and resource occupation conflicts,the study proposes a centralized,multi-to-multi task matching algorithm STCB to improve the stability.The existing task allocation mechanisms considering stability didn't consider the task conflict relationships,so in the experimental evaluation,we compare STCB with greedy algorithm,and the results show that STCB can match more users for tasks when the budget is insufficient,bring greater utilities for experimental users,meet the individual rationality and budget balance of experimental users and mobile users,and improve the stability of task allocation significantly.