On Compact Representation and Robustness of Transfer Patterns in Public Transportation Routing

Transfer pattern routing is a state-of-the-art routing algorithm for public transportation. It allows to answer multi-criteria shortest path queries within a few milliseconds. The algorithm is based on an expensive precomputation of shortest paths between stations. This thesis examines two aspects of transfer pattern routing. The first objective is to find a compact representation of the transfer patterns. We detect redundancy in the transfer patterns and present novel ways to store the information within less memory. The second objective is to analyze the robustness of the transfer patterns towards delay and to answer the question whether the patterns still allow for optimal routing, if the underlying network changes. We address the second question by introducing a framework to evaluate the robustness in networks with different delay scenarios and we present experimental results indicating that transfer patterns are quite robust.