MOOP: An Efficient Utility-Rich Route Planning Framework Over Two-Fold Time-Dependent Road Networks

Abstract

Utility-rich (e.g., more attractive or safer) route planning on city-scale road networks is a common yet time-consuming task. Although both travel time and utility on edges are time-dependent concurrently in real cases, they are overlooked in most prior work. In this paper, we focus on the route planning over two-fold time-dependent road networks, i.e., both travel time and utility on edges are varying over time. We aim to find a route from an origin to a destination by maximizing the accumulated utility score within a time budget. Moreover, to satisfy users' real-time requests, the fast response is usually mandatory. Here, we propose a novel two-phase framework called <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MOOP</b> , i.e., <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>M</u></b> anaging <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>O</u></b> ffline data for <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>O</u></b> nline route <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><u>P</u></b> lanning, to discover the near-optimal driving routes efficiently. Specifically, in the offline phase, we construct the auxiliary data structure, i.e., the edge table, to manage the time-dependent information about edges. In the following online phase, the route is generated sequentially by an iterative edge table visiting process. We evaluate the proposed <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MOOP</b> thoroughly based on synthetic road networks and two real-world road networks in the city of Chengdu (4,819 nodes and 6,385 edges) and the city of Chongqing (5,056 nodes and 7,355 edges) in China. Results show that: (i) our framework can work adaptively for different time-varying utility patterns; (ii) the edge table is economic yet effective; (iii) our route planning algorithm outperforms other baselines in obtaining the highest utility value while costing the least running time.