RYNSORD: a novel, decentralized algorithm for railway networks with "soft reservation"

Abstract

This paper presents RYNSORD (a novel decentralized algorithm with reservation for efficient scheduling and congestion mitigation in railway networks), wherein every train utilizes lookahead to dynamically re-plan its route. Thus, a train, currently at position X and headed for the ultimate destination position Y, first requests and reserves N tracks, ahead of its current position, for use at appropriate future times. The N tracks are chosen either along the shortest path or an alternate shortest path where the tracks of the two paths are mutually exclusive. The train issues two sets of requests, concurrently, for the tracks on the two paths. After it has been granted the train selects the path that promises the quickest arrival time at the final destination. The train moves through the N tracks and, upon completion, it again requests and reserves N subsequent tracks ahead of itself. The process continues until the train reaches its destination, Y. The reservation process may be characterized by less abruptness and greater flexibility in negotiation between the trains and stations. This characteristic is termed reservations in this paper. Under soft reservations, a train negotiating for N tracks will, in general, always succeeds. This paper also reports on an implementation of RYNSORD and a distributed approach that is similar to RYNSORD except that it employs hard on a network of SUN spare workstations, configured as a loosely-coupled parallel processor. Experimental results are presented.