Research On Energy Consumption Optimization Of Wireless Networks Based On ADMM And Q-Learning

With the emerging of 5G networks,mobile data traffic will grow in an explosive manner in the coming years,and the growth of traffic will increasingly drive the adoption of dense networks for wireless networks.The dense networking mode poses a great challenge to the energy consumption of the base station.This paper proposes a wireless network system for energy optimization based on Cloud Radio Access Network(C-RAN),aiming to reduce the number of active base stations and the consumption of transmit power,thereby reducing the overall energy consumption of the wireless network.The number of active base stations is related to the switching base station strategy,and the consumption of transmission power is related to the user association strategy.Therefore,this paper proposes a Q-learning-based and ADMM energy optimization scheme that considers both the switching base station and the user association.A new energy consumption optimization scheme based on KM algorithm is proposed for the switch of the base station.Furthermore,this paper mainly includes the following two aspects:(1)Energy consumption optimization strategy based on Q-learning and ADMM.This energy consumption optimization strategy considers the switching of base station and user association that are will be solved by different algorithms.In terms of the switching of base stations,Q-learning-based base station switch solutions are used to manage the number of base stations in the active state.With the guarantee of the quality of service,a small number of base stations are used to provide network services and the number of base station in the active state is reduced,thereby saving network energy consumption.On the other hand,when the user is associated with a base station with a poor signal,it will consume more transmit power to meet the user's demands.Therefore,this paper proposes an ADMM algorithm to solve the user association,so as to minimize the transmission power consumption and reduce wireless network energy consumption.(2)Energy consumption optimization strategy based on KM algorithm.This strategy mainly considers the energy consumption of the base station,starting with the switching of base stations.By considering the user and the base station as two sets of vertex of the bipartite graph,and establishing the weights between the two vertex sets.The matching of the weighted bipartite graph is done according to our extension and improvement algorithm of the KM algorithm.That is,the relationship between the user and the base station is established so as to achieve the purpose of using as few base stations as possible to satisfy the user service requirement.This strategy is also used to compare and verify the performance of Q-learning and ADMM based energy consumption optimization strategy.Based on the above research content,a lot of simulation experiments are presented to verify the performance of energy consumption optimization strategies based on Q-learning and ADMM,and energy optimization strategies based on KM algorithm.Because the KM algorithm-based strategy has energy optimization effects,so for the experiment more rich and more reliable,we use it to verify the performance of the main research point of this paper based on Q-learning and ADMM energy optimization strategy.