A new strategy for synchronizing traffic flow on a distributed simulation using SUMO

Abstract

The Project Modelling and Control of Urban Traffic in the City of Medell ́ın (MOY- COT) has produced multiple results in modelling, simulation and control of multimodal urban traffic using the SUMO simulator. As the simulations became more complex the necessity to distribute the computational load rose. Therefore, an approach for network partitioning and border edges management was introduced. In this paper a new border edge management strategy is presented for distributed simulation with SUMO. Unlike the previous approaches, which were developed in Python programming language using the corresponding TraCI client and tools such as sumolib, the strategy presented in this work was developed in C++ using the TraCI client for this language. Additionally, this strategy involves a simplified process for network partitioning since the border edges are preserved in every partition, without the need of splitting them. In this case, neighboring partitions behave in a master-slave fashion, depending on whether the border edge is an incoming edge or an outgoing edge. Concretely, a given partition is a master for its incoming edges and a slave for its outgoing ones. Furthermore, all the vehicles are found in the master and the slave partitions, where the master partition controls its slaves through the TraCI commands slowDown and moveTo that correct the position of these vehicles. Simulation results show that this new strategy presents better precision than the previous one. The description of the new procedure for border edge management is detailed. Finally, it is compared with the previous approach and the non-distributed simulation using a free flow scenario and a scenario with queue formation is presented.