Research On Joint Beam Allocation Algorithm Based On Multi-cell Interference Learning

With the continuous popularization of multimedia mobile communication equipment,the number of UEs has exploded,thereby increasing the interference between cells,which poses a greater challenge to the quality of data communication services.In high-speed mobile communication scenarios,the UE moves faster,and the beam allocation scheme at the previous moment may no longer be applicable,but if the beam allocation scheme is recalculated each time,it will undoubtedly increase system overhead.On the other hand,the traffic patterns in a fixed number of communities are highly similar in a fixed time period of the day and can be used.The millimeter wave beam itself is easily affected by the environment and multi-cell interference.Therefore,it is very challenging and necessary to propose and apply a robust and efficient millimeter wave beamforming technology.In recent years,computer hardware and computing power develop rapidly,AI and ML algorithms have entered the practical stage.ML algorithms can solve complex problems with less complexity and have attracted enough attention in the field of communication technology.This thesis first proposes a data-driven millimeter wave joint beam allocation algorithm based on multi-cell interference learning for static traffic patterns,which can reduce the impact of multi-cell interference on data communication services with less complexity.For dynamic traffic patterns,by using historical record similarity comparison,new sample selection algorithms,exploration-utilization and other strengthening mechanisms,the static traffic pattern-oriented algorithm can quickly and accurately assign millimeter wave beams to dynamic traffic patterns.Furthermore,a millimeter wave online joint beam assignment algorithm based on multi-cell interference learning is proposed.The proposed online algorithm can use past learning records to accelerate the convergence of the beam allocation scheme space,and finally give a near-optimal beam allocation scheme under the current traffic mode.In the comparative experiment,the proposed algorithm is compared with the random beam assignment method and UCB algorithm.The proposed algorithm converges faster and is more robust.It can provide a near-optimal beam allocation scheme with less complexity for mobile communications in high-speed mobile scenarios,and reduce the impact of multi-cell interference on system performance.