CIF: Small: Strategic Information Provision for the Control of Cyber-Socio-Physical Transportation Networks

现代交通网络是这样一个系统，它的正常运行取决于人、算法和物理因素之间的紧密相互作用，因此，它的监管（无论是为了减少拥堵还是鼓励使用节能模式）必须涉及所有组成部分，包括以人的决策为目标的机制。现有的方法通常要么依靠货币支付（例如通行费），要么依靠信息提供（例如使用可变信息标志或路由应用程序）来影响通勤者的选择。这两种机制之间的主要区别在于，前者是战略性的（因为货币支付/税收的计算和执行带有诱导特定社会有益行为的明确目标），而后者则不是（因为提供的信息只是对道路实际状态的报告）。该项目的目标是为信息提供机制的原则设计开发工具，以适应此类网络-社会-物理（CSP）系统的特定条件和约束，并积极地制作信息，使其与社会效率结果保持一致。本研究项目建立在贝叶斯说服和信息设计的现有框架上，并使用控制和信息理论的工具、观点和见解，将这些方法扩展到CSP系统的挑战性领域，特别是交通网络。具体而言，本项目提出了(1)在此背景下发生的非贝叶斯和有限理性决策的新模型，(2)非经典信息结构下的战略消息传递，并解决了相应的信息设计问题。该项目还(3)开发了必要的工具，将消息个性化作为引入信息解耦的手段，而不会在通勤者之间造成不同的影响。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Modern transportation networks are systems whose proper operation depends on tight interactions between human, algorithmic, and physical elements and whose regulation (be it to reduce congestion or encourage the use of energy-saving modalities) must thus accordingly involve all components, including mechanisms that target human decision-making. Existing approaches typically either rely on monetary payments (e.g., tolls) or information provision (using, e.g., variable message signs or routing apps) to influence commuters' choices. A major difference between these two types of mechanisms is that the former is strategic (since monetary payments/taxes are computed and implemented with the explicit goal of inducing a particular socially beneficial behavior), while the latter is not (since the information provided is simply a report of the actual state of the road). The goal of this project is to develop tools for the principled design of information provision mechanisms which are adapted to the specific conditions and constraints of such cyber-socio-physical (CSP) systems and actively craft messages to align them with socially efficient outcomes.This research project builds on the existing frameworks of Bayesian Persuasion and Information Design from Economics and uses Control and Information Theory tools, viewpoints, and insights to extend these approaches to the challenging realm of CSP systems in general, and traffic networks in particular. Specifically, this project formulates new models of (1) non-Bayesian and bounded rational decision making occurring in this context and, (2) strategic messaging under non-classical information structure and addresses the corresponding information design problems. The project also (3) develops the necessary tools to incorporate message personalization as a means of introducing informational decoupling without causing disparate impact among commuters.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.