Scalable Algorithms for Bicriterion Trip-Based Transit Routing

The paper introduces innovative techniques to enhance Trip-Based Transit Routing (TBTR), a popular bicriterion transit routing approach. Inspired by Hypergraph-based Round-based Public Transit Routing (HypRAPTOR), we present Hypergraph-based TBTR (HypTBTR), a partitioning variant aimed at improving TBTR’s query times. However, this improvement in HypTBTR (and HypRAPTOR) comes with increased preprocessing requirements. To address this issue, we propose two novel techniques: a One-To-Many variant of TBTR and multilevel partitioning. Our One-To-Many algorithm efficiently handles profile queries, and integrating multilevel graph partitioning in both HypTBTR and HypRAPTOR significantly reduces preprocessing computations. Extensive experiments on large-scale GTFS datasets of Switzerland, Netherlands, and Sweden, showcase the efficacy of the proposed algorithms. Empirically, HypTBTR outperforms TBTR by 23–37%, the One-To-Many variant shows a speed-up of 90–95%, and multilevel partitioning reduces preprocessing calculations by up to 53%. These improvements have the potential to enhance user experiences in web and mobile navigation applications and expedite problems such as transit assignments that involve repeated calls to routing algorithms.