Analytical and Scalable Analysis of Transient Tandem Markovian Finite Capacity Queueing Networks
提出一种解析模型近似瞬态串联有限容量排队网络的联合队列长度分布，复杂度与队列数线性相关且与容量无关，适用于大规模网络分析，并用于城市交通动态信号控制。
This paper proposes an analytical model to approximate the transient aggregate joint queue-length distribution of tandem finite (space) capacity Markovian networks. The methodology combines ideas from transient aggregation-disaggregation techniques as well as transient network decomposition methods. The complexity of the proposed method is linear in the number of queues and is independent of the space capacities of the individual queues. This makes it a suitable approach for the analysis of large-scale networks. The transient joint distributions are validated versus simulation estimates. The model is then used to describe urban traffic dynamics and to address a dynamic traffic signal control problem. The signal plan analysis shows the added value of using joint distributional information, and more generally spatial-temporal between-link dependency information, to enhance urban traffic operations. The online appendix is available at https://doi.org/10.1287/trsc.2015.0629 .