AG Kommunikationstheorie


A New Markov Chain Model for the Performance Analysis of Traffic Networks


In traffic networks like railway lines or roads, capacity is often limited by certain critical areas like junctions or single-tracked sections, which can be seen as servers with multiple queues for jobs arriving from different directions. The behaviour of these servers can be quite complex. For example, it might be possible to process jobs from several queues silmutaneously, but not for all possible combinations. Service times can be deterministic or random, and in both cases they might depend on which queue is served, or on some internal state of the server. This work presents an approach to model such a system by a Markov chain, with a multi-dimensional state space that is composed of the number of jobs in each queue, and possibly a finite number of internal states. Apart from the given constraints of the system (e.g., which queues can be served silmutaneously), the transitions of the Markov chain also describe the chosen service strategy. Thus, it is possible to compare the performance of different service strategies, as well as the performance of different systems (e.g., an existing level junction and a grade-separated new version of that junction)

zurück zur Terminübersicht