Martin Strehler

Traffic Signal Optimization Using Cyclically Expanded Networks

Traditionally, the coordination of multiple traffic signals and the traffic assignment problem in an urban street network are considered as two separate optimization problems. However, it is easy to see that the traffic assignment has an influence on the optimal signal coordination and, vice versa, a change in the signal coordination changes the optimal traffic assignment. In this paper we present a cyclically time-expanded network and a corresponding mixed integer linear programming formulation for simultaneously optimizing both the coordination of traffic signals and the traffic assignment in an urban street network. Although the new cyclically time-expanded network provides a model of both traffic and signals close to reality, it still has the advantage of a linear objective function. Using this model we compute optimized signal coordinations and traffic assignment on real-world street networks. To evaluate the practical relevance of the computed solutions we conduct extensive simulation experiments using two established traffic simulation tools that reveal the advantages of our model.

Routing of Electric Vehicles: Constrained Shortest Path Problems with Resource Recovering Nodes

We consider a constrained shortest path problem with the possibility to refill the resource at certain nodes. This problem is motivated by routing electric vehicles with a comparatively short cruising range due to the limited battery capacity. Thus, for longer distances the battery has to be recharged on the way. Furthermore, electric vehicles can recuperate energy during downhill drive. We extend the common constrained shortest path problem to arbitrary costs on edges and we allow regaining resources at the cost of higher travel time. We show that this yields not shortest paths but shortest walks that may contain an arbitrary number of cycles. We study the structure of optimal solutions and develop approximation algorithms for finding short walks under mild assumptions on charging functions. We also address a corresponding network flow problem that generalizes these walks.

Capacitated confluent flows: complexity and algorithms

A flow on a directed network is said to be confluent if the flow uses at most one outgoing arc at each node. Confluent flows arise naturally from destination-based routing. We study the Maximum Confluent Flow Problem (MaxConf) with a single commodity but multiple sources and sinks. Unlike previous results, we consider heterogeneous arc capacities. The supplies and demands of the sources and sinks can also be bounded. We give a pseudo-polynomial time algorithm and an FPTAS for graphs with constant treewidth. Somewhat surprisingly, MaxConf is NP-hard even on trees, so these algorithms are, in a sense, best possible. We also show that it is NP-complete to approximate MaxConf better than 3/2 on general graphs.

Traffic signal optimization using cyclically expanded networks

Traditionally, the coordination of multiple traffic signals and the traffic assignment problem in an urban street network are considered as two separate optimization problems. However, it is easy to see that the traffic assignment has an influence on the optimal signal coordination and, vice versa, a change in the signal coordination changes the optimal traffic assignment. In this article, we present a cyclically time-expanded network and a corresponding mixed integer linear programming formulation for simultaneously optimizing both the coordination of traffic signals and the traffic assignment in an urban street network. Although the new cyclically time-expanded network provides a model of both traffic and signals close to reality, it still has the advantage of a linear objective function. Using this model, we compute optimized signal coordinations and traffic assignment on real-world street networks. To evaluate the practical relevance of the computed solutions, we conduct extensive simulation experiments using two established traffic simulation tools that reveal the advantages of our model.

Polynomial-time algorithms for special cases of the maximum confluent flow problem

A flow on a directed network is said to be confluent if the flow uses at most one outgoing arc at each node. Confluent flows arise naturally in destination-based routing. We study the maximum confluent flow problem (MaxConf) with a single commodity but multiple sources and sinks and heterogeneous arc capacities. It was recently shown that MaxConf is NP-hard even on trees. We improve the classification of easy and hard confluent flow problems by providing polynomial-time algorithms for outerplanar graphs with a single sink, as well as trees with a constant number of either sources or sinks. Furthermore, we present an FPTAS for graphs with bounded treewidth.

Traffic Signal Optimization: Combining Static and Dynamic Models

In this paper, we present a cyclically time-expanded network model for simultaneous optimization of traffic assignment and traffic signal parameters, in particular offsets, split times, and phase orders. Since travel times are of great importance for developing realistic solutions for traffic assignment and traffic signal coordination in urban road networks, we perform an extensive analysis of the model. We show that a linear time-expanded model can reproduce realistic travel times especially for use with traffic signals and we verify this by simulation. Furthermore, we show how exact mathematical programming techniques can be used for optimizing the control of traffic signals. We provide computational results for real world instances and demonstrate the capabilities of the cyclically time-expanded by simulation results obtained with state-of-the-art traffic simulation tools.

Two FPTAS for the Constrained Shortest Path Problem Applied to Hybrid Vehicle Routing

We consider a constrained shortest path problem with two resources. These two resources can be converted into each other in a particular manner. Our practical application is the energy optimal routing of hybrid vehicles. Due to the possibility of converting fuel into electric energy this setting adds new characteristics and new combinatorial possibilities to the common constrained shortest path problem (CSP). We formulate the resulting problem as a generalization of CSP. We show that optimal paths in this model may contain cycles and we state conditions to prevent them. The main contribution is a polynomial-time approximation scheme and a simpler approximation algorithm for computing energy-optimal paths in graphs.

Optimizing Traffic Signal Timings for Mega Events

Most approaches for optimizing traffic signal timings deal with the daily traffic. However, there are a few occasional events like football matches or concerts of musicians that lead to exceptional traffic situations. Still, such events occur more or less regularly and place and time are known in advance. Hence, it is possible to anticipate such events with special signal timings. In this paper, we present an extension of a cyclically time-expanded network flow model and a corresponding mixed-integer linear programming formulation for simultaneously optimizing traffic signal timings and traffic assignment for such events. Besides the mathematical analysis of this approach, we demonstrate its capabilities by computing signal timings for a real world scenario.

The Maximum Flow Problem for Oriented Flows

In several applications of network flows, additional constraints have to be considered. In this paper, we study flows, where the flow particles have an orientation. For example, cargo containers with doors only on one side and train coaches with 1st and 2nd class compartments have such an orientation. If the end position has a mandatory orientation, not every path from source to sink is feasible for routing or additional transposition maneuvers have to be made. As a result, a source-sink path may visit a certain vertex several times. We describe structural properties of optimal solutions, determine the computational complexity, and present an approach for approximating such flows.