Controllability of transportation networks. (December 2018)

• Record Type: Journal Article
• Authors: Rinaldi, Marco
• Is Part Of: Transportation research. Volume 118(2018)
• Page Start: 381
• Page End: 406
• Publication Date: 2018-12
• Subjects: Network-wide traffic control -- Controllability -- Optimization; Transportation -- Research -- Periodicals; Transportation -- Mathematical models -- Periodicals
• DOI: 10.1016/j.trb.2018.11.005
• Languages: English
• ISSNs: 0191-2615

Abstract

Highlights: The control theoretical concept of controllability is reformulated to the specific instance of transportation networks; Modelling assumptions and requirements necessary to correctly capture traffic dynamics, both in terms of propagation and congestion and in terms of route choice are analysed; The Controller Location Problem is approached through an exact approach, generating solutions that guarantee full controllability; The relationship between amount and location of controllers and network-wide control policies' performances is investigated, in the specific instance of pricing controllers, through a series of case studies; The proposed control theoretical approach is compared with other controller location approaches both in terms of reachable optimality and of resilience to route choice information mismatch. Abstract: Network-wide traffic control policies determine the optimal values for the different kinds of controllers equipped on a transportation network, with the objective of reducing delays and congestion, improving safety and reaching a target Level of Service. While models and algorithms for these problems have been extensively studied in literature, little attention has been devoted to investigating whether/how different locations, kinds (pricing controllers, traffic lights, …) and amounts of controllers in a network affect the overall performance of network-wide schemes. In this work, we adapt the control-theoretical approach of controllability of complex networks to the specific instance of transportation networks, considering both propagation/spillback dynamics and users' behavior in terms of route choice. Thanks to the newly developed methodology, we then provide exact solutions to the Full Controllability Pricing Controller Location Problem for transportation networks. Comparing different pricing controller location policies through two artificial test cases, we empirically demonstrate how indeed the amount and kind of controllers in a network strongly affect the level of performance reachable by network-wide control policies, specifically in terms of Total Cost minimization.