Research On Task Allocation Of Spatial Crowdsourcing Based On Benefit Optimization

Despite the rapid development of science and technology,many working machines are still difficult to match human labor,especially when considering the difficulty and cost of equipment deployment and maintenance.With the development of the Internet,people-centric data collection methods have become very popular.The mobile crowdsourcing model makes full use of group intelligence to solve large-scale perception work that traditional computing methods cannot handle.By publishing tasks on the crowdsourcing platform,many users with smart devices decide whether to perform tasks according to their own wishes.Due to the different technical level of people,and they are difficult to manage,the reasonable assignment of tasks determines the quality of tasks,especially for some tasks with high time limit requirements,it is more required to select workers with strong working ability,and finally the publisher will give workers a certain reward.Therefore,for crowdsourced task publishers,reasonable task allocation and proper incentive mechanism are particularly important.In addition,due to the uncertainty of the task release time,the number of people on the crowdsourcing platform may not be sufficient to complete the task on time and with quality.Thus,the interaction between people in the social network needs to be considered,so that online people can spread the task to their social neighbors,and attract more people to the crowdsourcing platform to compete for tasks.Based on the above points,in order to improve data quality and maximize profits,an incentive mechanism is needed to encourage user participation.This paper first studies the reputation mechanism,participant selection,task allocation and joint pricing in mobile crowdsourcing systems.We propose a user reputation evaluation method and design a participant selection algorithm(PSA)based on user reputation.Then,we proposed a social welfare maximization algorithm(SWMA),which achieves task pricing that maximizes the interests of all parties,including task publishers and mobile users.Then,we divide the problem of social welfare maximization into local optimization sub-problems which can be solved by double decomposition.Through extensive simulations,it is proved that the SWMA algorithm converges to the optimal solution,which proves the algorithms PSA and SWMA are effective.For the issue of the small number of online users in the crowdsourcing platform when task publishing,we propose a task diffusion framework.The diffusion cost estimated by the framework,the results of the task diffusion and the number of dynamically changing task publishing make the publisher choose whether to continue the diffusion strategy.Furthermore,we design winner selection in task diffusion algorithm(WSTSA)and winner reward algorithm(WRA)based on user influence in social networks.Experiments on real social network datasets prove the effectiveness of the algorithms WSTSA and WRA.