| Crowd intelligence provides a new way of using the power of crowds and clusters of computing devices to collect,process,predict,and detect large amounts of useful information.Among them,the task publisher issues a large number of tasks to a large number of non-professional workers through the crowd intelligence platform in order to obtain reliable answers.A crowd intel-ligence system includes three stakeholders,namely the platform,the workers(two modes of existence,including humans and computing devices),and the requester(including both humans and machines).This article aims at the existing problems of the traditional crowd intelli-gence,such as lack of incentive mechanisms,high operating costs,poor user experience,and protection of user data privacy.By combining with the hard-ware infrastructure and network environment of mobile edge computing,and the benefits of blockchain smart contract technology,we can build a set of soft-ware and hardware systems for a decentralized edge-enabled crowd intelligence platform.Our contributions are:1)It shows that a crowdsourcing community involves three stakeholders,namely the requester,the worker,and the crowdsourcing platform,and the in-terests of the three conflict with each other.We propose and test a hypothesis of the crowdsourcing community that all workers believe that in most cases,they find that the true answer to each crowdsourcing task is only disturbed by un-biased noise,leading to the generation of false answers,and A crowdsourcing mechanism is designed,which uses a series of reward and punishment func-tion pairs and workers' individual rank values to unify the interests of different stakeholders.These advantages have been confirmed by theoretical analysis and subsequent simulation experiments.Our work helps to free the platform and publishers from the efforts of monitoring workers and selecting capable workers,and only cares about solving the crowdsourcing task itself,saving the cost of publishers and attracting more Professional workers work on the plat-form.2)The traditional cloud computing center is replaced by the edge cloud network of mobile edge computing to realize its decentralized deployment and management operations.Using the relationship between the crowdsourcing completion time,the number of workers,and the number of characters,a spe-cially designed commission mechanism model is introduced to ensure that the master node honestly intercepts a certain number of tasks to itself,and trans-fers the remaining tasks to adjacent master nodes to achieve The distribution of tasks between master nodes.Specifically,each master node calculates the number of tasks it should intercept based on the deadline set by the publisher,the number of online workers in the edge cloud service area of the master node,and the estimation of the quality of the workers.If there are remaining tasks,they are forwarded to other nearby edge clouds until all tasks are distributed.Note that during the distribution process,the master node can only see the hash tag of the task,but not the specific information of the task.This is to ensure the randomness of the task distribution.The distribution process is public on the blockchain.3)In order to better support decentralized applications,collaboration be-tween edge clouds is particularly important.In order to achieve seamless mi-gration,path selection is proposed to find one or more paths to transfer service data from the source edge server to the target edge server.We pay attention to the following issues of path selection:how to deploy,and whether its con-trol is centralized or decentralized;because mobile users and network providers have different interests in path selection,how to balance the two;How to en-sure seamless service migration when the migration time must be less than a certain threshold.This work is the first to consider the importance of path se-lection in service migration,and use the traffic guidance module of 5G edge computing to implement service migration.This work proposes a path selec-tion algorithm that can optimize both the network level and the service level.The problem is modeled as a multi-objective optimization problem,and it is theoretically proved that the proposed algorithm can give a weak Pareto opti-mal solution.Not only that,in order to improve the scalability of the proposed algorithm,a distance-based filtering strategy algorithm was designed to filter out unnecessary switches in advance.4)The geographically distributed edge cloud network provides the infras-tructures,where the edge cloud can provide the computing,storage,and net-work resources required for the platform to operate.Blockchain technology is used to solve the trust problem of collaboration scenarios between edge clouds.The design of the platform involves the evaluation of crowdsourced answers,the calculation of final answers,the determination of functions such as workers-compensation and rewards due to submitted answers,and the business processes that must be considered to achieve decentralized trust.The R&D language of the entire platform is friendly to the application of the Internet of Things,and we has completed the initial global node deployment and it runs smoothly. |
PDF Full Text Request