Mobile Crowd Sensing Task Allocation Based On Improved Fireworks Algorith

Mobile crowdsensing is a new mode of data acquisition that combines crowdsourcing ideas and mobile device sensing capabilities.It takes advantage of the distribution and mobility of mobile users,and uses the smart devices carried by the user and the existing communication infrastructure to form a large-scale data-aware network.Task allocation is an important part of mobile crowdsensing system.A reasonable allocation scheme can balance the interests of service requesters,platforms and participants under the multiple constraints of time,skills and workload.Fireworks algorithm is a meta-heuristic optimization algorithm with simple structure and strong global optimization ability.It is suitable for solving NP-hard problems such as mobile crowdsensing task allocation.Based on the above background,this paper studies the model and method of mobile crowd sensing task allocation based on improved fireworks algorithm.The research contents are as follows:Firstly,considering the psychological and behavioral processes of participants,the mathematical model of heterogeneous task allocation problem in mobile crowdsensing is established by introducing environmental information,participants’ health status,credibility and measurement time.In order to solve the model,a predictive discrete fireworks algorithm is proposed.The algorithm uses the distance and measurement time in the model to design the fireworks explosion operator;the grouping linear prediction strategy of explosion amplitude and the adaptive competition mechanism of mutation operator are introduced.The performance of the improved strategy and the proposed algorithm is verified by one mobile crowdsensing heterogeneous task assignment instance and 10 synthetic instances.Compared with six representative algorithms,the proposed algorithm can search for task allocation schemes with less total platform cost.Secondly,for the platform to publish multiple projects,each project contains multiple heterogeneous tasks,a mobile crowdsensing heterogeneous multi-project multi-task allocation model is established.The model completes the project in a group collaboration mode,distinguishes the functions of members in the group,and considers the impact of credibility and skill level on the sensing quality of the project.In order to solve the model,an ensemble multiobjective fireworks algorithm is proposed.The algorithm introduces a dual-feedback ensemble learning framework;four weak optimizers are designed according to the explosion characteristics of fireworks algorithm.The weight of weak optimizer is adjusted by the feedback of evolutionary significance.The individual evaluation mechanism is corrected based on the feedback of the difference in objective exploration.The comparison results with five representative algorithms on 12 instances show that the proposed algorithm can provide a set of Pareto non-dominated task allocation schemes with better convergence and distribution for the platform.Finally,focusing on the dynamic characteristics of tasks and participants,a mathematical model of online multi-project multi-task allocation based on trajectory prediction is established.Two dynamic events are introduced,a sparse region compensation mechanism is added,and a semi-Markov discrete-time model is used to predict participant trajectories.In order to solve the model,the multi-objective fireworks algorithm is used to generate the task allocation scheme at the initial time.In the process of system execution,in order to quickly respond to dynamic events occurring in each cycle,a neighborhood search algorithm based on ensemble learning is proposed.The change response mechanism is designed based on the transition probability.Through the framework of ensemble learning,the neighborhood search operator that is most suitable for the current environmental characteristics is selected independently.The comparison results with four online task allocation algorithms on 16 instances show that the proposed algorithm can plan the allocation scheme with higher sensing quality and lower platform cost at each scheduling time.