Continuous Predictive Line Queries for On-the-Go Traffic Estimation

Traffic condition is one vital piece of information that any commuter would wish to obtain to plan an efficient route. However, most existing works monitor and report only current traffic, which makes it too late for commuters to change their routes when they realize they are already stuck in the traffic. Therefore, in this paper, we propose a traffic prediction approach by defining and solving a novel continuous predictive line query. The continuous predictive line query aims to accurately estimate traffic conditions in the near future based on current movement of vehicles on the roads, and continuously update the predicted traffic conditions as vehicles move. The predicted traffic condition will not only help redirect commuters in advance but also help relieve the overall traffic congestion problem. We have proposed three algorithms to answer the query and carried out both theoretical and empirical study. Our experimental results demonstrate the effectiveness and efficiency of our approach.