Dynamic braess's paradox in complex communication networks

Braess's paradox shows a counterintuitive scenario where more resources may cause worse traffic performance when all users pursue their personal optimum. Previous study on this topic mainly focused on the specific traffic volume to show the existence of such a paradox. For real-world complex communication networks, on the contrary, the traffic volume changes from time to time. Therefore, it is important to study the traffic performance under this dynamic condition. We find that Braess's paradox does not happen when the traffic volume is sufficiently low or sufficiently high. Between these two extremes, the probability that Braess's paradox happens also changes with the traffic volume. This result is helpful for communication network planning and management.