Algorithms for most reliable routes on stochastic and time-dependent networks

Abstract

This study presents algorithms to determine the most reliable routes on stochastic and time-dependent networks. The measure of reliability adopted is the probability of on-time arrival at the destination, given a threshold arrival-time. We propose two distinct algorithms to determine optimal time-adaptive strategy and optimal apriori path on stochastic and time-dependent networks. First, a decreasing order-of-time algorithm is proposed to determine the optimal strategy to the sink from all node and departure-time combinations. Second, a label-correcting, network pruning algorithm is proposed to determine the optimal path between the source and the sink for a given departure-time. The correctness of both the proposed algorithms is proved and their computational complexity expressions are derived. The efficacy of the proposed procedures is demonstrated on large-scale transportation networks. This work has the potential to facilitate wider application of stochastic and time-dependent networks in reliability-based modeling and analysis.