Research On Channel Estimation And Resource Management For 5G WTTx

Wireless To The x(WTTx),as a high-speed,multi-link,and wide-coverage wireless access technology,has once again achieved rapid development in 5G systems.The performance of the physical layer channel estimation will directly determine the effectiveness and reliability of the WTTx system,but the improvement of the throughput and error performance of most WTTx systems is severely restricted by actual scenarios,so appropriate physical layer design and channel estimation are necessary to ensure the flexible deployment of WTTx.On the other hand,the current research on WTTx not only has a single specific business scenario,but there are few documents considering the wireless resource management in WTTx.However,resource allocation requires frequent data exchange,which affects user privacy and security and leads to leakage of sensitive information.At the same time,privacy protection is one of the core requirements of user-centric 5G.Federated learning can realize multi-user data sharing under the premise of protecting user privacy.Therefore,resource management based on federated learning can not only avoid privacy leakage,but also allocate wireless resources reasonably to maximize the energy efficiency of the WTTx system.We propose the following solutions and innovations for the above two problems.Firstly,we design a new WTTx time-domain frame structure as the basis for the design of new characteristics of the physical layer and the study of channel estimation algorithms.We propose the new frame structure as follows.1)Introducing additional pilots;2)Sharing Demodulation Reference Signal(DMRS)inter-time slots;3)Desiging non-orthogonal port interference estimation receiver.The introduction of additional pilots improves the ability to combat channel fading and is conducive to obtain more accurate channel response.Through transfering the channel estimation value among time slots,the sharing of DMRS among time slots can save pilot overhead and enlarge transmission resources of the Physical Downlink Shared Channel(PDSCH).The interference between quasi-orthogonal DMRS ports are considered in desingning onorthogonal port interference estimation receiver,and original signal are regained based on ways to cancel interference.Simulations verify that the introduction of additional pilots greatly reduces the block error rate of the system,DMRS sharing significantly improves system throughput,and the non-orthogonal port interference estimation receiver improves the estimation accuracy and reduces the error probability of the system.Secondly,considering the privacy leakage problem faced by data sharing in the WTTx scenario,we propose a WTTx-based federated learning framework,and deeply explores the resource management issues in federated learning.We proposes a data set partitioning algorithm based on data importance selection.By selecting important data instead of all data for local calculation of the terminal,on the one hand,it can significantly reduce the calculation energy consumption and delay,on the other hand,the risk of privacy leakage is greatly reduced.Thirdly,we set the optimization goals of the resource management problem based on the importance of data as learning efficiency and total system energy consumption,and optimize these two parts to obtain the optimal data set size and bandwidth resource allocation ratio.The multi-objective optimization problem is solved with the help of the Non-dominated Sorting Genetic Algorithm-?(NSGA-?).The simulation results prove that the algorithm significantly improves the model convergence rate and greatly reduces the system energy consumption.