Joint Optimization Of Channel Allocation And Power Control In WLAN

In recent years,with the rapid development of wireless communication technology,and smart terminals and Io T devices have been widely popularized in ordinary people's homes.To satisfy users' network requirements,wireless access points have been thickly deployed in various places.However,high-density APs have serious interference,which impact the user's experience,resulting in low network throughput and poor connection quality.At the same time,in dense WLAN,it is difficult to estimate the dynamic change of the network and the burden of real-time calculation is large.Therefore,the traditional optimization algorithm is difficult to solve the interference problem in dense scenarios.To solve the above problems,this thesis proposed a joint optimization algorithm of channel allocation and power control based on reinforcement learning,which utilizes the centralized control idea of SDWN technology.Through interactive learning between the agent and the environment,the algorithm obtains an optimization strategy for jointly adjusting the AP channel and power,thereby system interference is reduced and throughput is improved.First of all,through the deployment of actual measurement experiments,the upper limit of AP throughput performance is measured,and the correlation between the AP's transmit power,working channel,and system throughput is clarified.In addition,a mathematical optimization model for joint adjustment of AP transmit power and channel with throughput as the optimization target is established.Secondly,according to the system throughput performance optimization goal,the power and channel state of the AP is converted into a state space for reinforcement learning,the channel and power adjustment strategy of the AP is an action space,and the system throughput increment that can guide the autonomous learning of the agent is a reward function of reinforcement learning.Then we obtain the optimal strategy by adjusting the parameters in offline training.In addition,to adapt to the dynamics of largescale networks,a joint optimization mechanism based on event-driven Q-learning is designed.The mechanism triggers learning through event conditions and periodically updates the optimal strategy,thereby effectively reducing the resource consumption caused by algorithm operations.Finally,the performance of the algorithm is verified through simulation.The experimental results show that the joint learning optimization algorithm based on reinforcement learning can effectively improve the throughput while ensuring the user's communication requirements.Compared with the existing channel and power joint optimization algorithm,the optimization algorithm proposed in this thesis can increase the system throughput by up to 9%,while the system interference is reduced by up to 21%.