Decentralized model predictive control of freeway traffic systems over lossy communication networks

In this work, we present a decentralized scheme for the control of a freeway traffic system over a non-ideal trans-mission channel, involving network delays ultimately resulting in possible packet losses in the digital communication. The feed-back scheme exploits the well-known Cell Transmission Model (CTM) and a ramp metering control, with actuation consisting of on-ramp traffic lights. In this context, the Model Predictive Control (MPC) approach is applied locally, in a decentralized way, which is preferable from the computational viewpoint with respect to centralized solutions, the complexity of which would grow rapidly with the dimension of the system at hand, making real-time traffic control potentially ineffective. The simulation setup is based on a traffic model whose parameters are identified using data produced by a commercial microscopic simulator of the traffic system. Preliminary performance results show the effectiveness of the approach taken, also in the case of parameter uncertainties, and look promising in view of future investigations.