| Millimeter wave(mmWave)and large-scale multiple input and multiple output(MIMO)technology can effectively improve the communication performance.However,the high directivity of the beam generated in a large-scale MIMO system is narrow and the need for frequent beam switching in high-speed motion scenarios leads to a significant increase in mmWave system overhead and beam misalignment probability.Beam tracking technology can effectively reduce the system overhead and ensure the accuracy of the beam direction,thus improving the performance of the whole communication system,especially in highspeed motion scenarios.Therefore,this thesis focuses on the beam tracking problem in millimeter wave massive MIMO systems in motion scenarios.Existing beam tracking algorithms based on motion scenarios usually use a simplified single linear uniform motion to build a nonlinear system model,which cannot be directly applied to the complex and common curved road conditions in real production activities.Therefore,this thesis proposes a millimeter wave beam tracking algorithm based on curved trajectory,which establishes a universal curved trajectory model with selected position,velocity and yaw angle parameters,which can be applied to a variety of motion trajectories and overcomes the dependence of Bayesian filter-based beam tracking algorithm on the model;since curved trajectories are more unpredictable and highly susceptible to sudden disturbance noise,the tracking accuracy is degraded in the volume Since the curve trajectory is more unpredictable and prone to degradation of tracking accuracy due to sudden disturbance noise,a graded correction factor is introduced on the basis of Cubature Kalman Filter(CKF)algorithm to correct the observed error covariance matrix and the weight of the predicted values to overcome the influence of sudden disturbance values.The simulation results show that in the general complex curve trajectory scenario,the improved CKF algorithm improves the tracking accuracy compared with the classical Extended Kalman Filter(EKF)algorithm,the more accurate Unseented Kalman Filter(UKF)algorithm and the original CKF algorithm by 7.57 dB,2.33 dB and 1.29 dB,respectively;in the special linear trajectory scenario,the tracking accuracy is improved by 10.43 dB.Further,in order to address the shortcomings of the existing literature that only a single specific driving behavior of the vehicle user is studied in the same scene,this thesis proposes a millimeter wave beam tracking algorithm for the existence of multiple driving behaviors of the vehicle user,which adopts the Interacting Multiple Model(IMM)theory to establish a motion scene with multiple driving behaviors.Secondly,to solve the beam misalignment problem caused by the rapid and abrupt change of beam angle due to the continuous switching between driving behaviors,we propose an improved Interacting Multiple Model-Adaptive Kalman Filter(IMM-Adaptive Kalman Filter)algorithm by adaptively correcting the weight of the prediction parameter yaw angle change rate using a priori position information and transfer probability.Cubature Kalman Filter,IMM-ACKF)to reduce the probability of beam misalignment.The simulation results show that the improved IMM-ACKF algorithm improves the tracking accuracy by 20%compared with the IMM-ACKF algorithm in the case of dealing with abrupt behavioral state changes;at SNR=0dB,the improved IMM-ACKF algorithm improves the tracking accuracy by 10%at maximum compared with the IMM-ACKF algorithm. |
