Improving Large Scale Real-Time Ridesharing With Heuristics For Road Networks

Despite the great potential, ridesharing, especially with dynamic queries, has not been thoughly investigated on large-scale road networks often with many turns and one-way roads. In this paper, we treat one-way roads and viaduct roads of urban road network as heuristic information, and improve the response time of a two-phased ridesharing framework by pre-pruning some drivers to avoid complex shortest path calculation. We evaluate the proposed method using the data sets, which consist of road network and the large-scale trajectory data respectively generated by Brinkhoff road network generator in the city of Jinan, China. Results show that, our algorithm achieves a very high pruning ratio with little response time-only about 40ms in the case of 5000 drivers. It can be concluded that the improved framework is more effective for solving practical application problems.