Traffic Management in Freeway Networks with Unreliable Link Capacities

My PhD research focuses on traffic management in freeway networks subject to random disturbances that compromise link capacities. Such disturbances are responsible for a significant portion of congestion in transportation networks, which leads to enormous economic and environmental losses. My research develops a framework that systematically considers the presence of unreliable capacities in the assessment and design of traffic management strategies. Specifically, I model link capacities as variables that stochastically switch between a finite set of values. Two types of traffic management capabilities are considered, viz. routing (network level) and ramp metering (link level). At the network level, I combine the stochastic capacity model and the deterministic queueing model, and develop a novel piecewise-deterministic queueing (PDQ) model, which quantifies the network-wide impact of unreliable capacities. At the link level (ramp metering), I combine the capacity model with the cell transmission model, which results in a stochastic cell transmission model (SCTM). Based on the theory on convergence of Markov processes, I obtained results on stability and steady-state distribution of the PDQ model and the SCTM. I also explored the control design problem by formulating and solving a PDQ network optimization problem. My ongoing work is in three directions. First, I am working on calibrating the stochastic capacity model using real traffic data. Second, I am investigating the consistency between the PDQ model and the SCTM. Third, I am synthesizing the network level and the link level to design joint routing-ramp metering policies.