We study the problem of controlling phase activations for signalized junctions in an urban transportation network using local feedback information consisting of measures of the queue-lengths at the incoming lanes of each junction. Our focus is on the validation and performance evaluation through micro-simulations of the recently proposed Generalized Proportional Allocation (GPA) controller. Previous theoretical work has provided provable performance guarantees in terms of stability, and throughput optimality of the GPA controller in a continuous averaged dynamical queueing network model. In this paper, we first provide and implement two discretized versions of the GPA controller in the SUMO micro simulator. We then compare, in an artificial Manhattan-like grid, the performance of the GPA controller with those of the MaxPressure controller, which is another distributed feedback controller that requires more information than the GPA. Finally, to show that the GPA controller is easily implementable in a real-world scenario, we apply it to a previously published realistic traffic scenario for the city of Luxembourg and compare its performance with the static controller provided with the scenario as well as with the cyclic MaxPressure controller. The simulations show that the GPA controller outperforms both the fixed time and the cyclic MaxPressure controllers for the Luxembourg scenario, and behaves better than the MaxPressure pressure controller in the Manhattan-grid when the demands are low.

中文翻译：

---

广义比例分配交通信号控制的微观仿真研究

我们使用本地反馈信息研究控制城市交通网络中信号交叉口的相位激活问题，该信息包括每个交叉口进入车道的队列长度的测量。我们的重点是通过最近提出的广义比例分配 (GPA) 控制器的微观模拟进行验证和性能评估。先前的理论工作在连续平均动态排队网络模型中 GPA 控制器的稳定性和吞吐量优化方面提供了可证明的性能保证。在本文中，我们首先在 SUMO 微模拟器中提供并实现了 GPA 控制器的两个离散版本。然后，我们在人工曼哈顿式网格中比较 GPA 控制器与 MaxPressure 控制器的性能，这是另一个需要比 GPA 更多信息的分布式反馈控制器。最后，为了表明 GPA 控制器在实际场景中易于实现，我们将其应用于先前发布的卢森堡市现实交通场景，并将其性能与场景中提供的静态控制器以及循环 MaxPressure 控制器。模拟表明，GPA 控制器在卢森堡场景中的性能优于固定时间和循环 MaxPressure 控制器，并且在需求较低时比曼哈顿网格中的 MaxPressure 压力控制器表现更好。我们将其应用于先前发布的卢森堡市真实交通场景，并将其性能与场景中提供的静态控制器以及循环 MaxPressure 控制器进行比较。模拟表明，GPA 控制器在卢森堡场景中的性能优于固定时间和循环 MaxPressure 控制器，并且在需求较低时比曼哈顿网格中的 MaxPressure 压力控制器表现更好。我们将其应用于先前发布的卢森堡市真实交通场景，并将其性能与场景中提供的静态控制器以及循环 MaxPressure 控制器进行比较。模拟表明，GPA 控制器在卢森堡场景中的性能优于固定时间和循环 MaxPressure 控制器，并且在需求较低时比曼哈顿网格中的 MaxPressure 压力控制器表现更好。