Research On Wi-Fi Error Control Methods Based On Machine Learning

With the increasing number and throughput of Wi-Fi connections,Wi-Fi puts forward higher technical requirements which will also become the main driving force for the development.However,the transmission unreliability and delay instability are the main challenges faced to the current Wi-Fi technology.The IEEE 802.11 ax standard aims to improve the throughput and transmission reliability in dense scenarios where Wi-Fi signals interfere with each other severely and the frequency band resources are tight due to a huge mount of accesses,which leads to more prominent retransmission problems.In the draft process of the next-generation IEEE802.11 be standard,high throughput is mostly taken as the main design point,and there are currently few proposals to solve the reliability of Wi-Fi transmission.In the existing study on Wi-Fi retransmission,more theoretical studies on retransmission have been carried out,including Wi-Fi retransmission schemes,HARQ theoretical efficiency and coding schemes,etc,which less consider the retransmission combined with IEEE 802.11 standards.The Wi-Fi retransmission study carried out in this thesis aims to design a new retransmission solution based on the new features of the existing IEEE 802.11 ax standard and the future IEEE 802.11 be standard.The specific contributions of this thesis are as follows:First of all,this thesis designs two new types of retransmission schemes,namely multiband HARQ intelligent retransmission based on the existing Wi-Fi6 standard IEEE 802.11 ax and joint retransmission between APs in Multi-AP scenarios based on the next-generation WiFi7 standard IEEE 802.11 be.For multi-band HARQ intelligent retransmission based on IEEE802.11 ax,this thesis combines the characteristics of multi-band and HARQ with 802.11 ax based on scheduling communication.Three retransmission strategies are proposed for different network scenarios: insisting on the current frequency band(ICFB)retransmission,switching to the backup frequency band(SBFB)retransmission and concurrent retransmitting on dual-band(CRDB).The three retransmission modes can be adapted to different network conditions.For multi-AP joint retransmission based on IEEE 802.11 be,a multi-AP network topology scenario is constructed and two retransmission schemes for Overlap Basic Service Set Area(OBSSA)are designed,which are respectively the optimization of the retransmission environment for OBSSA interference coordination and the best neighbor AP assisted retransmission.According to the characteristics of the interaction between APs,the communication process and signaling frame structure between APs are designed,which mainly include the collection and reporting of channel state information(CSI),the feedback of coordination results,and the coordination request and response of neighbor APs Frames etc.Then,in terms of algorithm design,for the multi-band retransmission scheme,the twolayer algorithm model of offline unsupervised clustering and online supervised deep learning is designed.According to the current spectrum resource status,channel SINR value and channel SINR value of each band and HARQ outage rate,the best retransmission mode is determined.In this algorithm,the clustering classifies and calibrates the network status,and calibrates each network status applicable to different retransmission modes to form a data set.The deep learning process trains a neural network based on the data set of the offline process to obtain an online neural network classifier.Joining the advantages of the two algorithms,the algorithm avoids the overhead caused by the high complexity of the clustering algorithm and uses an efficient neural network to find the optimal retransmission mode,which effectively realizes that the periodic update of the offline clustering data set and online neural network efficiently chooses the the best retransmission mode.For the OBSSA retransmission scheme,interference coordination algorithm based on genetic algorithm(GA)optimizes the STA's resource and power allocation in OBSSA and the best neighbor AP selection algorithm based on the deep Q-learning network(DQN)selects the best neighbor AP.Compared with the traditional exhaustive algorithm and random selection algorithm,the algorithm designed in this thesis has higher efficiency and accuracy in the application of OBSSA's retransmission scheme.Finally,the theoretical analysis and simulation results show that the designed retransmission schemes can improve the Wi-Fi retransmission efficiency and data transmission reliability,and the scheme based on IEEE 802.11 ax can control the average transmission delay within 10 ms,and the average throughput is greatly improved compared with the traditional retransmission scheme.The retransmission of OBSSA also enables the edge STAs to have higher communication capabilities and stability,which makes the Wi-Fi transmission support more network use cases to meet the requirements of Wi-Fi reliability and stability.