Research And Implementation Of High-precision Positioning In Ad Hoc Network Based On Millimeter Wave Radar

The vehicular ad hoc network(VANET)is a distributed mobile network composed of a group of vehicle nodes with limited communication and computing capabilities in traffic scenarios.It aims to realize the interconnection between different vehicles,vehicles and roadside fixed access points,vehicles and people,while providing services such as accident warning,assisted driving and Internet access.VANET is considered to be a key technology to realize smart transportation,promote the industrialization of smart cars,as well as improve the efficiency of urban operation.Node’s location is the basic of the related theories and network applications.Positioning in the complicated traffic scenarios is also an important and difficult problem for VANET.This article relies on the National Natural Science Foundation Innovation Group Project(Project Approval Number:61821001)and the National Natural Science Foundation Key Project(Project Approval Number:61531007)to conduct research on the high precision positioning of VANET in traffic scenarios.The paper summarizes the existing wireless network node positioning methods,comprehensive positioning accuracy,system stability,driving safety,technological advancement,future applications and other factors considered,the thesis adopt millimeter wave radar to achieve the goal of VANET node positioning.The contributions and innovations of the thesis mainly include the following aspects:1)Established the functional framework of millimeter wave radar for VANET positioning.Solve the three main problems of radar module and communication module fusion positioning:problem of node positioning blind area,problem of multi-node radar data fusion processing,problem of radar detection target and communication node mismatch,and clarify the relationship between node communication range and radar detection range.2)Propose an algorithm for matching and positioning neighbor nodes in VANET.The matching function is established from the measured data of the radar modules between two adjacent nodes,the distance dimension and angle dimension are introduced to characterize the degree of matching.Based on matching function,the VANET node positioning technology is realized.The positioning technology has the characteristics of distributed,self-organization and strong stability,while the positioning error can reach the centimeter level.The simulation researches the influence of radar measurement error and the total number of nodes on the matching positioning error rate and positioning error,and conducts off-site experiments to verify the scheme.3)Proposed two high-precision positioning algorithms.Both algorithms use the radar measurement data of multiple neighbor nodes.One establishes an optimization least square equation aimed at minimizing the location error;Considering that the influence of angle estimation error on positioning will increase with the neighbor’s distance increase.Another multi-neighbor node distance-weighted positioning algorithm is given.This algorithm gives higher weight to the near neighbor node.The simulation results show that the two algorithms can effectively improve the node positioning accuracy.In comparison,the least squares method is more sensitive to the number of nodes,and the distance weighting method has higher accuracy and more stable performance under the same conditions.Finally,an off-site experiment was performed to compare and verify the two algorithms.4)Realize the positioning and tracking technology of millimeter wave radar for multi-vehicle targets.This is the key technology to achieve thesis design objectives.The thesis divides the radar signal processing into three parts:front-end,point cloud,and target,and realizes basic point cloud technologies such as ranging,speed measurement,and angle estimation.Under the framework of Kalman filtering,the relationship between the radar detection and the vehicle motion model is established,and the vehicle target recognition and tracking algorithm is proposed.The effectiveness of the algorithm is verified by hardware-in-the-loop simulation,and pedestrians and obstacles can be distinguished.