Research On The Key Technology Of 5g And Wifi Coexistence In High-frequency Unlicesed Band

The transformation of wireless communication is the core of the fourth industrial revolution,and 5G communication technology brings the Industrial Internet of Things into the Industry 4.0 era.With the deployment of billions devices in the industrial network scenario,the massive data communication makes the problem of spectrum scarcity more and more serious in wireless communication systems.To realize real-time,reliable,and efficient wireless communication between industrial equipment,5G has begun to adopt high-frequency unlicensed millimeter wave bands for data transmission owing to its abundant spectrum resources.As an important part of the high-frequency unlicensed band,the unlicensed 60GHz millimeter wave spectrum has attracted widespread attention from industry and academia due to its large bandwidth and unlicensed characteristics.Considering that the IEEE 802.11 ad and IEEE 802.11ay networks have been deployed in this band,the key point to the effective deployment of 5G is to realize the coexistence and resource management of 5G and WiFi networks in non-cooperation mode and cooperation mode.In view of this,to fulfill the demand of high-throughput and lowlatency in the Industrial Internet of Things,this thesis focuses on fair channel access,beam switching and resource allocation problems in noncooperative coexistence mode,and coordinated data transmission technology of 5G and WiFi systems in cooperative coexistence mode,which is closely combining the millimeter-wave narrow beam directional transmission characteristics.Based on nonlinear fitting,convex optimization,and machine learning methods,this thesis proposes a directional LBT aided random access mechanism,trajectory prediction and channel monitoring aided fast beam switching mechanism,joint user association,beam selection and power allocation resource management algorithms,and data offloading algorithms based on deep reinforcement learning.The proposed scheme avoids interference between heterogeneous network devices,reduces the latency and signaling interaction overhead,improves the spectrum efficiency of 5G network,and reduces the corresponding user service latency,which enables a friendly coexistence of 5G and WiFi systems.The main achievements and contributions of this thesis are summarized as follows:(1)First of all,this thesis focuses on the high latency problem caused by the high complexity of narrow beam search in directional LBT access and beam switching,and conducts research on fair access and fast beam switching mechanisms under non-cooperative coexistence mode.Due to the application of Massive MIMO and directional narrow beams,the complexity of beam alignment increases exponentially,which brings great difficulties to the initial directional LBT channel access process in static scenarios and the beam switching in dynamic scenarios.Therefore,based on the single-user multi-link Massive MIMO system,this thesis proposes a joint directional LBT and beam training channel access mechanism for the establishment of the initial communication link in a static scenario.On this basis,a trajectory prediction and channel monitoring aided fast beam switching algorithm is proposed for mobile scenarios to reduce the latency and signaling interaction overhead during channel access and beam switching,and improve the quality of service of user communication links.The simulation results show that the proposed channel access mechanism has an average increase of 31%compared with the comparison algorithm in terms of sum rate,and reduces the complexity of beam pair link to 1/3.From the perspective of latency,the results show that the latency of the proposed fast beam switching scheme is reduced by an order of magnitude(2)Secondly,considering the difficulty and complexity of resource allocation and interference avoiding in 5G systems on account of the introduction of beams,this thesis studies the resource management scheme of 5G systems under non-cooperative high-frequency coexistence mode.Although the use of directional beams can reduce inter-link interference and improve space division multiplexing significantly,the interference between heterogeneous networks still exists and cannot be ignored at this time.In addition,since beam selection,user association and power control are independent optimization processes,the optimal beam pair selected by the 5G device may not be the optimal communication beam pair in the heterogeneous network.To maximize spectrum efficiency of the 5G network while ensuring a friendly coexistence,we formulate an optimization problem by jointly considering beam selection,user association and power control.More specifically,we design a spectrum planning mechanism to reduce the interference between 5G and WiFi,and then a block coordinate descent method is used to determine the user association,beam selection,and power control for 5G users,while limiting the interferences to WiFi devices.The simulation results show that the proposed algorithm can converge in 3 iterations.Through the comparison of performance,it can be found that the successive convex approximate power allocation algorithm makes up the shortcomings of the greedy user association algorithm,which can achieve almost the same spectral efficiency of heuristics convex approximation algorithm with lower computational complexity.(3)Finally,considering that the total service delay is unacceptable when unidirectional data offloading is adopt,this thesis studies the coordinated transmission mechanism under cooperative coexistence mode.Since the number of users on a single 5G BS is limited by the number of RF links,the data packet can only queue for free spectrum resources when the number of users exceeds the load capacity of 5G network.As a result,the average service delay of users increases.On the other hand,because the WiFi network employs the CSMA/CA mechanism to acquire channels access permission,it can only serve one user at a same time.In massive machine type communication scenario,the collision probability of the data packets increases as the number of user increases,the data transmission delay becomes larger.In order to meet the low-latency transmission requirements of a large number of industrial equipment,we employ aggregation node in WiFi network to aggregate data from multiple users in parallel,and then the aggregation data is sent to the AP side through the CSMA/CA mechanism.On this basis,we formulate an optimization problem to minimize the average service delay of users under heterogeneous coexistence scenarios,and a deep reinforcement learning based bi-directional data offloading algorithm is proposed to improve the data transmission rate of heterogeneous networks.On top of that,the heterogeneous networks can achieve a win-win situation for coexisting.The simulation results show that the Q value of the proposed algorithm can be converged after 100 iterations,and the user service latency and the throughput of heterogeneous network are very close to that of the traditional convex optimization algorithm.