Resource Scheduling Strategy Based On Hybrid Intelligent Reinforcement Learning Automata In Fog Computing

Fog computing is an emerging cloud computing paradigm that is closer to users,more geographically distributed,and widely used by mobile users.Resource scheduling in fog computing is a challenging multi-objective optimization problem.Many optimization algorithms with good performance have been successfully applied in resource scheduling of cloud computing and achieved remarkable results.In recent years,fog computing has gradually been paid close attention to by the academic circle,and the industry has also started to research and practice platforms to conduct innovative research on fog computing platform.Therefore,combining learning automata model with fog computing environment is crucial to the research on resource scheduling strategy.With the development of artificial intelligence,resource scheduling needs to cater to the machine learning trend of big data,which makes it have the characteristics of digitization,communication,cooperation,intelligence and sustainability.For resource scheduling in the Internet of things(IoT),we minimize end-to-end latency,achieve load balancing,and reduce user and service provider costs.Firstly,a user-fog-cloud model based on multi-agent method is proposed,which can learn from the environment and adopts the proposed hierarchical structure Q-learning automata(HQLA)resource scheduling strategy.A hierarchical Q-learning automata model(HQLA)is proposed for training and learning model parameters where HQLA resource scheduling strategy in fog computing is implemented by top-down service publishing and bottom-up user resource scheduling.In particular,the top-down HQLA service publishing algorithm and bottom-up HQLA user resource scheduling algorithm are presented.The multi-objective optimization problem of fog resource scheduling is transformed into a single-objective optimization problem by weighted linear regression.A large number of experimental results show that our resource scheduling strategy reduces user latency,improves the execution efficiency of the system under the condition of better load balancing,and reduces the cost of users and fog service providers.Then,the competition mechanism of fog computing service providers based on Stackelberg game is proposed.The competition model between fog service providers is established,and the influence between investment scale and profit of different fog service providers is discussed under the condition of asymmetric market position in fog computing environment.The Stackelberg fog service provider competition mechanism model(SFCM)is proposed,under the Stackelberg model,the ratio of output between two service providers is 2:1.Furthermore,by introducing the entry cost into SFCM,an extended Stackelberg fog service provider competition mechanism model(ESFCM)is proposed.The influence of entry cost on the strategies of both parties is discussed through the Stackelberg model of Stackelberg competition.It is concluded that the mainstream large fog service providers with large market share have great competitive advantages and first-mover advantages.Finally,the game reinforcement learning automaton(GRLA)resource scheduling strategy in fog computing environment is proposed.In the cooperative game analysis center(CGAC)of fog computing node,the alliance of users is carried out through user election,and cooperative game is carried out among users.This paper proposes a game reinforcement learning automata model(GRLA),which preprocesses the data by statistical analysis and management of the types and quantities of tasks,and adopts both the Sarsa algorithm and the results of cooperative games for fog computing resource scheduling.The user selection algorithm and sorting selection algorithm are proposed to select the representative users with cooperative potential,then the GRLA based resource scheduling algorithm in fog computing is proposed.Extensive experimental results show that the GRLA based resource scheduling strategy in fog.computing improves the overall execution speed of the system,eliminates resource waste at the root,and reduces the cost of users and fog service providers.