On-line map-matching framework for floating car data with low sampling rate in urban road networks

The performance of map matching has a significant effect on obtaining real-time traffic information. The floating car data (FCD) is of low-sampling rate, and urban road networks such as multi-layer roads can be particularly complex. Most of the current low-sampling-rate map-matching approaches use a fixed time interval, which can result in a lack of efficiency and accuracy if the initial point is not correctly matched. Moreover, the issue of handling data relating to multi-layer road networks remains open. To address these issues, a new on-line map-matching framework is proposed, comprising the confidence point and the maximum delay constraint dynamic time window. The framework performs map matching by self-adaptively choosing the appropriate timing and matching method according to the complexity of the local network to which the positioning point belongs. To distinguish elevated roads from normal roads, vehicle behaviour patterns on elevated roads are taken into account. Comparisons of the proposed algorithm, hidden Markov model algorithm, incremental algorithm and point-to-curve algorithm are conducted on two datasets. The empirical results show that the proposed algorithm outperforms the other algorithms. When the behaviour pattern on elevated roads is considered, the accuracy of these algorithms is also improved.