Improved Map Matching Algorithm Based On C-Measure And Its Spark Implementation

With the improvement of people's living standard,automobile consumption has grown rapidly.How to manage automobiles comprehensively in urban environment is an urgent problem to be solved.Map Matching,as the core technology of intelligent transportation system,can effectively reduce errors in positioning,as it can provide service for the intelligent transportation system by determining the best matching location of the vehicle.The traditional map matching algorithm fails to strike a balance between accuracy of vehicle location and the efficiency of service calculation,an improved map matching algorithm ICMM(Improved C-Measure Map Matching)is proposed.On the one hand,based on the traditional C-Measure measurement method,many factors and weighting mechanisms are designed,it includes four factors: the distance between vehicle position and the line,the average distance of historical track and the line,the direction deviation of travel direction of the vehicle position and that of the candidate line,as well as deviation between travel direction of the continuous positions and that of the candidate line section.Weight coefficients relevant to above-mentioned factors are also included in the design.The experimental results confirm that this method can effectively improve the location accuracy of the map matching algorithm,the complexity will not increase significantly.On the other hand,in the process of algorithm implementation,a new candidate line segment selection mechanism is adopted to reduce the candidate line range and horizontal adaptive fuzzy network(HAFN)is used to realize dynamic solution to parameters.Compared with the traditional algorithm,it can improve the stability of vehicle operation,but also ensure that the computational efficiency is not significantly reduced.The traditional map matching algorithm based on serial execution can't meet the large-scale data processing requirements,a parallel design idea is proposed,which is implemented by Spark technology and Kudu technology.Firstly,the improved map matching algorithm(ICMM)is designed in parallel on the Spark framework to meet the needs of large-scale data processing.Hilbert space-filling curve is adopted to create spatial index and then partitioned data is computed.Consequently,computing power on each node is balanced and the computing efficiency of the algorithm is improved.