A Context-aware Map Matching Method Based on Supported Degree

Map matching technology is indispensable in the tide of building smart cities. The difficulty degree of matching depends on the matching context (road network density and GPS point quality). However, most existing map matching algorithms use same strategies under different matching contexts, which are hard to balance accuracy and efficiency. Therefore, we propose a new method of map matching, which includes two matching phases: projection matching (Strategy 1) for every GPS points and connectivity matching (Strategy 2) for portions without credible results from the first phase. Thereinto, supported degree is employed to judge the credibility of the projection matching result, which reflect the difficulty degree of matching in each region. In the connectivity matching phase, for matching complex portions, tree structure is creatively adopted, which can represent the connectivity between roads. Besides, we present novel tricks to increase the efficiency, such as considering road segment as the basic element of map matching and simplifying connectivity tree based on limiting-velocity. Finally, to evaluate the performance of this new method, we have compared it with conventional algorithms on the same dataset, which consists of 480,973 GPS points. The proportion of the error road segment length in total trajectory length is used as the criterion to estimate the matching accuracy of the algorithm. When the sampling period is 10s, the algorithm can improve the matching accuracy rate to over 95%. Meanwhile, the running efficiency of this algorithm is obviously better than other algorithms, in sampling period less than 100s.