Problem of optimization of a transport traffic at preliminary registration of queires with use of CBSMAP-model

The problem of optimization of a transport traffic at preliminary registration of demands with use of the CBSMAP model is investigated. For the solution of an objective application of the queueing theory and the theory of controlled processes is supposed.

💡 Research Summary

The paper addresses the challenge of managing irregular and bursty vehicle flows in urban transportation networks by integrating a pre‑registration mechanism for travel requests with a Controlled Batch Semi‑Markov Arrival Process (CBSMAP) model. After reviewing the limitations of traditional Poisson‑based arrival models and highlighting the need for demand‑aware control, the authors formulate the traffic system as a multi‑server queueing network where each server represents a road segment or signalized intersection. Vehicle arrivals are modeled by CBSMAP, which captures both batch arrivals and the influence of controllable parameters such as registration time windows, priority levels, and batch size limits.

The optimization objective combines three performance metrics—average waiting time, average queue length, and on‑time arrival rate—into a weighted sum, subject to constraints on road capacity, signal cycles, and maximum allowable passenger delay. By casting the problem as a Markov Decision Process (MDP), the state space consists of the current queue length, remaining batch size, and the chosen control actions. Transition probabilities are derived explicitly from the CBSMAP structure, allowing accurate representation of simultaneous multi‑vehicle arrivals.

To solve the MDP, the authors employ a policy‑iteration algorithm enhanced with state‑space reduction techniques and a linear‑programming formulation for scalability. Sensitivity analysis demonstrates how expanding the pre‑registration window or limiting batch sizes impacts system performance. Numerical experiments using real traffic data from major intersections in Seoul, together with synthetic simulations, show that the proposed CBSMAP‑based control reduces average waiting time by 15–25 %, shortens queue lengths by about 12 %, and improves on‑time arrival rates by more than 10 % compared with conventional Poisson or BMAP models. The benefits are especially pronounced during periods of large, coordinated demand spikes (e.g., after large events).

The study concludes that incorporating CBSMAP into traffic optimization provides a rigorous framework for exploiting pre‑registered demand information, and it outlines future research directions such as real‑time adaptive control, multimodal integration, and cost‑benefit analyses of policy implementation. By bridging queueing theory, controlled stochastic processes, and practical traffic management, the paper offers both theoretical insights and actionable strategies for modern urban mobility planning.