Travel time prediction for Origin-Destination pairs without route specification in urban network

Travel time prediction is one of the important traffic information for both individual travelers and road system authorities. So, many researchers have developed several techniques for travel time prediction. However, most of the existing methods have been focusing on predicting travel time on highways, whereas developing methods particularly for urban road networks is still at premature stage. In this study, we propose a travel time prediction method for urban road network based on the concept of k Nearest Neighbors (k-NN). The proposed method differs from the existing methods in three aspects. First, it can predict the travel time for all Origin-Destination pairs in an entire urban road network. Second, it can effectively predict the travel time without considerations on complex factors of a road network, such as queuing delay, signal process, and path assignment. Third, it shows the robust accuracy even in low percentage of detector coverage. Lastly, for the application of the proposed k-NN method to the real-world traffic information service, we investigate the key design parameters.