Physics-inspired strategies for understanding and optimizing transportation networks

Tak Shing TAI

Research output: ThesisDoctoral Theses


Traffic congestion, which is caused by the increasing number of vehicles, occurs almost daily in urban areas. Owing to the limited space for road expansion, coordinated traffic routing is a better alternative to uncoordinated routing for reducing traffic congestion. In this study, we investigate the characteristics of traffic problems such as phase transitions from the free-flow state to the congested state and the effective dimension of systems, develop methods with random moves in routing to coordinate traffic flow and coordinate traffic flow in network with road blockage. We apply randomness in routing in a two-dimensional cellular automata model to try to sacrifice traveling distance of vehicles to maintain the free-flow state in transportation systems, which can maximize the movement count of vehicles. We then compare two different routing strategies, namely centralized/global routing strategies and individual/adaptive routing strategies, based on the randomness in routing in the model, to maximize the arrival count in the transportation systems. The results show that centralized routing strategies and less greedy routing strategies can be applied to denser networks to reduce traffic congestion. In other words, increasing the randomness in the routing can alleviate traffic congestion problems in denser networks. In addition, in our studies, the effective dimension of transportation systems is reduced from the perspective of physics when traffic congestion occurs. We also employ the cavity approach to obtain optimal diverted routes in networks with road blockage, where vehicles travel from different origins to a common destination. The proposed optimization algorithms for traffic diversion are tested in simulated networks such as random regular graphs with different connectivity and square lattices, and real networks to reveal their effectiveness. When some broken links exist in the networks accidentally, which block the connections between nodes, the results show that the influence of road blockage reduces with the coordination of traffic diversion because the increase in traveling cost can be suppressed. The increase in the connectivity of networks can also reduce the influence of road blockage since it increases the alternative routes for diversion. The coordinated traffic diversions are tested in UK highway networks and the results show that the increase in the traveling cost can be suppressed by up to 66%, compared with the diversions with shortest paths. These studies would help improve traffic coordination in the future, particularly with full driving automation. All rights reserved.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • The Education University of Hong Kong
  • YEUNG, Chi Ho 楊志豪, Supervisor
  • YEUNG, Yau Yuen, Supervisor
  • POON, Kin Man, Supervisor
Publication statusPublished - 2021


  • Statistical physics
  • Cellular automata
  • Randomness in routing
  • Cavity method
  • Traffic diversion
  • Theses and Dissertations
  • Thesis (Ph.D.)--The Education University of Hong Kong, 2021.


Dive into the research topics of 'Physics-inspired strategies for understanding and optimizing transportation networks'. Together they form a unique fingerprint.