Data-driven road side unit location optimization for connected-autonomous-vehicle-based intersection control

Low communication delay is crucial for the effectiveness of connected-autonomous-vehicle-based (CAV-based) intersection control strategies. To achieve low vehicle-to-road-side-unit (V2R) communication delay and support the implementation of CAV-based intersection control strategies, this study addresses the problem of road side unit (RSU) location optimization at a single intersection. Considering the uncertainty of the selection of intersection control strategies, the problem is formulated as a two-stage stochastic mixed-integer nonlinear program. The model aims to minimize the sum of the cost associated with RSU investment and the expectation of the penalty cost associated with V2R communication delay exceeding a pre-determined threshold. The first stage of the program determines the number and location of RSUs, when the intersection control strategy to be implemented is unknown. Given the first stage decision and the implemented intersection control strategy, the second stage model optimizes the detection area allocation among RSUs to minimize the penalty cost. The model is linearized using the piecewise linearization technique. Then an integer L-Shaped algorithm is proposed to find a global optimal solution to the linearized program. In the numerical example, the proposed model is compared with a deterministic model. The results demonstrate that the V2R communication reduction per cost obtained by the proposed model is 28.95 larger than that obtained by the deterministic model, in the scenario that a CAV-based control strategy is implemented in the second stage. This indicates that the proposed model provides cost-effective low V2R communication delay for intersection control in CAV environment.