高速车载多天线自适应波束成形定位技术研究

Traditional beamforming algorithms have serious challenges in the accuracy of vehicle antennas positioning and channel estimation with high speed. In this proposal, we try to researching multi-antennas adaptive beamforming positioning algorithm based on separate receive and training antennas (SRTA) for very fast moving vehicle. Project Mainly solve the problems of vehicle received SNR decreased and position mispointing between receiving antennas and predictor antennas in three aspects including vehicle antennas mispointing, vehicle antennas receiving signal-to-noise ratio, AOA estimation of predictor antenna transmitted signals. We propose beamforming algorithm which adaptive switch off or active part of transmission antennas of BS based on SRTA, which can adaptive adjust beam width and coverage, with the aim to meet the needs of vehicle antennas received SNR. SRTA with polynomial interpolation channel estimation beamforming algorithm is presented, which can accurately estimate the channel parameters. Then, a fast moving vehicles multi-antennas adaptive beamforming algorithm has been explored. In this scheme, algorithm can improve vehicle received SNR and decrease or eliminate beamforming mispointing. Try to researching AOA dynamic beamforming algorithm to estimate the direction and angle of predictor antenna signals. In this scheme, we combine it with SRTA polynomial interpolation algorithm and part of transmission antennas adaptive switch off and active beamforming algorithm of BS based on SRTA to improve vehicle receiving SNR and eliminating vehicle multi-antennas beamforming mispointing.

传统波束成形算法在高速车载天线定位和信道估计的准确性方面遇到严重挑战。本课题探索研究基于接收天线与训练天线分离（SRTA）的高速车载多天线自适应波束成形定位算法，着重从天线定位错位、天线接收信噪比、预测天线信号AOA估计三个方面考虑解决波束成形技术带来的接收天线与预测天线位置的定位错位及接收信噪比下降问题。给出基于SRTA的BS端部分传输天线自适应关闭和激活的波束成形算法，自适应调整波束宽度及覆盖面，满足车载接收天线接收信噪比需求。提出基于SRTA的多项式插值信道估计波束成形算法，精确估计信道参数，在此基础上设计高速车载多天线自适应波束成形算法，改善车载天线接收信噪比，以消除或减小车载天线定位错位。探索研究AOA动态波束成形算法估计预测天线来波方向和角度，辅助多项式插值SRTA算法和BS端部分传输天线自适应关闭和激活的SRTA算法，改善接收信噪比、消除车载天线波束成形定位错位。

传统波束成形算法在高速车载天线定位和波达方向估计的准确性方面遇到严重挑战。本项目主要研究了高速车载多天线通信中的波达方向估计和定位技术等关键问题，包括：1）研究了互质阵列空洞中内插阵元的DOA估计算法，针对互质阵列差联合构造的虚拟阵元中空洞不可用而导致阵列最大可用自由度减少的问题，提出了一种新的互质阵列中添加天线的方法。算法中内插最少的天线数使虚拟阵列达到最大可利用自由度，且内插天线数并不改变互质阵列的原始孔径大小，在相同信噪比情况下，实现更为精准的波达方向估计。针对波达方向（Direction of Arrival，DOA）估计中传统子空间算法需要进行特征值分解或奇异值分解等复杂计算问题，提出一种基于双平行线阵（Double parallel linear array）的DPLA算法，结果表明，算法避开了特征值分解或奇异值分解，降低了计算复杂度，提高了运算速度。2）针对车载通信网络的定位研究，给出实时长短时记忆（LSTM）递归神经网络（RNN）来跟踪和预测GPS定位误差方法，利用深度学习技术来提高准确度的方法提高GPS定位精度，同时在美国中部和东部几个州的城市和大都市、城市和高速公路上采集的大量实验数据进一步路测验证所提出的定位预测模型。所提出的实时LSTM的预测精度可在地面真值的1-3%以内，优于传统统计和线性预测模型的预测结果。3）针对提高车载定位GPS精度的算法需要对每个误差源进行复杂的分析建模，提出了一种基于长短时记忆（LSTM）的GPS误差预测方法。实验结果表明，整体预测准确率提高了16%，在较远的预测步提高了37.7%。4）基于车对车（V2V）通信的低传输延迟和丢包率（PDR）研究，开发研究建立了一个车对车（V2V）通信测试平台。选择几个典型的天气和道路条件下对平台进行验证实测，测试车辆之间距离相对较近的道路安全应用上，范围从1米到20米不等。实验结果表明，只有DSRC能够满足安全应用的所有时延要求，而LTE-4G网络通信在丢包率和覆盖率方面优于DSRC。