Map Matching For Urban Road Networks And Its Parallelization Method Research

Map matching is the process of determining the current location of a moving object by matching GPS track points to the corresponding road sections of the actual road network.The driving trajectory composed of a series of GPS trajectory points can explore the average change pattern on the total number of rows,but more importantly,it can analyze the spatial aggregation process formed by all individuals traveling.However,due to the influence of external noise,network transmission delay and other factors,the trajectory data collected by the location-aware equipment will inevitably deviate from the actual driving trajectory.Therefore,map matching algorithm is needed to match the GPS trajectory data to the correct road section in the road network.Map matching has important application value for location-based services in intelligent transportation systems,such as autonomous driving,location prediction,and path recommendation.By mining and analyzing the matched trajectory data,it can identify congested road sections,understand urban traffic conditions,and realize road load balancing,etc.Facing the complex road network environment,the map matching algorithm faces challenges in handling low-frequency GPS trajectory data,and there is room for improvement.Therefore,improving the accuracy and robustness of map matching algorithm becomes an urgent problem to be solved.Firstly,a map matching method based on Hidden Markov Filter(HMF)is proposed for low-frequency sampled GPS track data in urban road network environment,which effectively improves the accuracy of map matching.Secondly,in order to improve the efficiency of map matching,a floating grid road segment retrieval method is proposed.Compared with the existing methods,the proposed method can effectively balance the retrieval efficiency of track point candidate road segments and the map matching accuracy.Finally,a parallelized map matching model in distributed environment is proposed,which can effectively improve the map matching efficiency of large-scale trajectory data while ensuring the map matching accuracy.The specific research of this paper is as follows:(1)Data cleaning rules are formulated for the characteristics of abnormal data appearing in the trajectory data collected by the location-aware devices,and data cleaning is performed on the abnormal trajectory data.Secondly,in order to improve the map matching efficiency,the urban road network is gridded according to the GeoHash coding rules.(2)The existing track point candidate road segment retrieval method cannot simultaneously take into account the efficiency of track point candidate road segment retrieval and the accuracy of map matching.A floating grid-based road segment retrieval method is proposed.Firstly,based on the urban road network divided by GeoHash code,a floating GeoHash grid is used to retrieve candidate road sections for trajectory points more flexibly and efficiently.Secondly,the method is applied to the map matching method based on Hidden Markov Model,and the results show that the method can effectively solve the problems in the existing methods for retrieving candidate road sections for track points.(3)An HMF-based map matching method is proposed,which firstly introduces the information of the angle formed between the heading angle of the moving object and the direction angle of the road section in the calculation of the observation probability.Secondly,in the calculation of transfer probability,not only the time factor,which is more concerned by travelers,is taken as the influence factor,but also the historical transfer probability,which can reflect the popularity of different driving paths,is added.Finally,HMF is used as the map matching model,which effectively improves the map matching accuracy in the low frequency sampling environment.(4)Based on the map matching method of HMF,a parallelized map matching model in a distributed environment is proposed.The model performs map matching for multiple mobile objects’ trajectory data in a Spark distributed cluster environment.Finally,the performance of the parallelized map matching model is evaluated.This paper addresses the problem that the existing map matching methods have poor matching effect when facing the low-frequency GPS trajectory data in the complex road network environment.The experimental results show that the proposed method has higher matching accuracy and shows stronger robustness in the face of sparse trajectory data.Secondly,the parallelized map matching model in distributed environment effectively improves the map matching efficiency for large-scale trajectory data.