Scalable Algorithms for Bicriterion Trip-Based Transit Routing

Autor: Agarwal, Prateek, Rambha, Tarun
Rok vydání: 2021
Předmět:
Druh dokumentu: Working Paper
Popis: This paper proposes multiple extensions to the popular bicriterion transit routing approach -- Trip-Based Transit Routing (TBTR). Specifically, building on the premise of the HypRAPTOR algorithm, we first extend TBTR to its partitioning variant -- HypTBTR. However, the improvement in query times of HyTBTR over TBTR comes at the cost of increased preprocessing. To counter this issue, two new techniques are proposed -- a One-To-Many variant of TBTR and multilevel partitioning. Our One-To-Many algorithm can rapidly solve profile queries, which not only reduces the preprocessing time for HypTBTR, but can also aid other popular approaches such as HypRAPTOR. Next, we integrate a multilevel graph partitioning paradigm in HypTBTR and HypRAPTOR to reduce the fill-in computations. The efficacy of the proposed algorithms is extensively tested on real-world large-scale datasets. Additional analysis studying the effect of hypergraph partitioning tools (hMETIS, KaHyPar, and an integer program) along with different weighting schemes is also presented.
Databáze: arXiv