Research On Collaborative Work Algorithms For Fog Computing Nodes

Along with Internet-of-Things(IoT)gradually stepping into social production and life,people put forward more and more higer demands for IoT applications,like delay and bandwidth.In order to meet the requirements of real-time IoT applications,the fog computing network came into being.It overcomes the defects of limited bandwidth resources and long delay of the cloud computing center,and can quickly respond to the request of IoT devices without taking up a lot of network bandwidth.At present,the collaborative algorithm between fog computing nodes(FCNs)has become one of the research hotspots in the field.As a distributed system,fog computing network has high requirements for collaborative work algorithms,including fairness,scalability,stability and reliability.To meet the requirements of collaborative work algorithms for FCNs,the work and contributions made in this paper are as follows:(1)this paper analyzes the queuing model of FCNs,demonstrates the fairness of time-sharing mechanism(TS)by comparing the widely used First-Come-First-Serve mechanism(FCFS)and TS mechanism,and proposes the M/G/1 queuing system of TS as the queuing model of FCNs.Futhermore,the parallel coefficient of FCNs is proposed based on the fairness of queuing model of FCNs,which can objectively measure the workload of FCNs with multitype tasks.Then combining with the work stealing mechanism of the distributed system,the collaborative work algorithm of FCNs based on load balancing is proposed which is named LBWS.LBWS utilizes Nash bargaining solution to distribute the probability set fairly,so that the parallel coefficient of each FCN reach the Pareto optimal status.The comparative experiments of several parameters verify that the algorithm is fair to multiclass tasks and scalable to networks with different sizes.(2)this paper defines the relative reputation coordinates of FCNs in the evidence space and evaluation space.The trust,distrust and uncertainty based on uncertainty probability theory are derived by combining the task stealing process,and then evidence fusion algorithm and time discount algorithm are proposed.Futhermore,this paper designs the reputation model of FCNs and the reputation system for fog computing network.By Combining the reputation system and load balancing algorithm,a collaborative work algorithm of FCNs based on load balancing and reputation system is proposed which can work more efficiently and overcome the defect of performance fluctuation,which is named LBWSRS.In addition,relevant experiments demonstrate that this algorithm can alleviate the influence of performance fluctuation of FCNs and improve the reliability and stability of fog computing network.In the end,the next step of fog computing research is proposed,and the future is planned and prospected.This paper uses 24 figures,12 tables,54 references.