Sina Bahrami

The growing deployment of connected and automated vehicles (CAVs) creates new opportunities and challenges at the intersection of traffic engineering and cybersecurity. While CAVs use onboard sensors and computing to navigate their environment, the addition of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication is expected to improve cooperation, safety, and efficiency. This connectivity, however, introduces system-level vulnerabilities that are not present in isolated autonomous vehicles. In this paper, we develop a queueing-based analytical framework to examine how application-layer flooding attacks affect the real-time processing of detection and control signals within a CAV. In the model, connectivity carries a concrete operational value. The safety margin a vehicle must maintain narrows as the rate of trustworthy safety information grows, in line with the shorter gaps sustained by cooperative adaptive cruise control. By jointly optimizing the division of limited computational capacity between detection and decision-making tasks, together with the fraction of V2V traffic admitted into the pipeline, we derive closed-form relationships between processing rates, admission levels, and macroscopic traffic variables such as delay, flow, and speed. Under the model's processing-cost and safety conditions, the analysis characterizes the value of connectivity under attack. Below a critical malicious share, a vehicle optimally admits and processes V2V data, which raises flow and extends the operable density range beyond what onboard sensing alone can sustain. Above the threshold, the optimal defense is to disconnect from the V2V channel. When safety regulations preclude disconnection, we obtain in closed form a maximal sustainable attack intensity beyond which no feasible operating point remains. Our findings make the case for cybersecurity-aware traffic flow models and identify when admitting V2V data is worthwhile and when disconnection is the better defense.

On-demand delivery platforms coordinate customers, suppliers, and drivers in real time. Because an order is prepared while its driver is still traveling to the pickup point, the customer's waiting time depends on the supplier's queue as well as on pickup and delivery travel. We study this three-sided problem with endogenous participation and queue-dependent waiting times. In each decision epoch, the platform sets a customer delivery-fee rate, a supplier commission rate, and a driver wage rate, and dispatches drivers to accepted orders. The platform does not impose matches directly: a transaction forms only if the customer chooses the supplier from the available menu and the served supplier and driver each earn nonnegative utility, a structure we encode with binary availability variables linked to prices and waiting times. The resulting problem is a mixed-integer nonlinear program. Our solution alternates between two tractable steps, a linearized queue-aware matching problem at fixed prices and a linear pricing problem at fixed matches, with an ordering--delivering implementation that scales the matching step to larger instances. Experiments on synthetic urban instances compare this integrated control with fixed-price, initial-price, and queue-blind benchmarks. Integrated control earns the highest profit, and driver time costs emerge as the main constraint on service volume, altering both the prices the platform sets and the routes it can profitably serve. The results indicate that pricing, matching, participation, and queue-dependent waiting times should be optimized as one system rather than sequentially.

Integrating charging ports into existing parking lots offers a practical pathway to expand electric vehicle (EV) charging, but charger-equipped spots are typically reserved for EVs, creating two inefficiencies: conventional vehicles (CVs) are turned away while charger-equipped spots sit idle, and EVs occupy chargers after charging at no cost, blocking other EVs and worsening range anxiety. This study develops a joint sizing and pricing framework for retrofitting parking lots under shared access to charger-equipped spots. The operator chooses the share of spots to equip with chargers and sets three prices: a regular parking price, a charging price, and a blocking fee levied on any vehicle that occupies a charger-equipped spot without actively charging, while CV and EV users with heterogeneous dwell and charging times choose among normal spots, charger-equipped spots, and an outside option. We formulate the problem as a mathematical program with equilibrium constraints and, by characterizing the indifference dwell and charging times, reduce the equilibrium to closed-form polygon integrals. This yields the existence and uniqueness of the user equilibrium and a closed-form characterization of the profit-maximizing design, and it supports a numerical evaluation under alternative operator objectives such as EV charger access, and parking access parity between CVs and EVs. Benchmarking against current practice shows that the blocking fee, by deterring wasteful overstay, drives the gains in EV access and operator profit but the net welfare improvement comes from opening the spots to all vehicles, while profit--equity and profit--access frontiers quantify the trade-offs the operator faces. The framework offers policy insight on the equitable and efficient expansion of EV charging infrastructure.

Flood evacuation plans prioritize discharging regions that are more vulnerable to getting flooded. More people can evacuate by delaying the flood's arrival at vulnerable regions through optimal placement and control of the water flow from the reservoirs. This paper proposes an optimization model to jointly control the reservoir system (in a water network) and the evacuation plan (in a transportation network) for maximizing the number of evacuees. We present the traffic flow as a cell transmission model and the water flow as a Muskingum-Cunge flood-routing model. We formulate the problem as a nonlinear program. To solve this problem, we solve a linear relaxation of the model using generalized Benders decomposition. The Town of High River in Canada is chosen as a case study. We show that the integrated flood management system regulates the reservoir discharge rates and evacuation rates to minimize the impact of the flood.

Recent studies show that energy consumption of battery electric vehicles (BEVs) increases in traffic congestion. Therefore, it is important to consider the effect of link flow on BEV energy consumption. The flow-dependent energy consumption changes the route choice and user equilibrium conditions. In this paper, some shortcomings of available BEV flow-dependent energy consumption user equilibrium models are shown first. Then, “sufficient” as well as “sufficient and necessary” user equilibrium based on the generalized travel time of each path and sub-path penalties are defined and modeled for flow-dependent energy consumption. While it is difficult to solve the sufficient and necessary model, the sufficient model can be solved directly with commercial solvers for small to medium-sized networks by generating all paths. An iterative algorithm is also presented to generate paths as required to solve the problem for larger networks. Numerical examples demonstrate the model and proposed algorithm, and analyze the impact of flow-dependent energy consumption on equilibrium conditions.