Traffic Signal Optimization

Dynamic Traffic Assignment (DTA) provides an approach to determine the optimal path and/or
departure time based on the transportation network characteristics and user behavior (e.g., selfish or
social). In the literature, most of the contributions study DTA problems without including traffic
signal control in the framework. The few contributions that report signal control models are either
mixed-integer or non-linear formulations and computationally intractable. The only continuous linear
signal control method presented in the literature is the Cycle-length Same as Discrete Time-interval
(CSDT) control scheme. This model entails trade-off between cycle-length and cell-length.
Furthermore, this approach compromises accuracy and usability of the solutions.
In this study, we propose a novel signal control model namely, Signal Control with Realistic Cyclelength
(SCRC) which overcomes the trade-off between cycle-length and cell-length and strikes a
balance between complexity and accuracy. The underlying idea of this model is to use a different time
scale for the cycle-length. This time scale can be set to any multiple of the time slot of the Dynamic
Network Loading (DNL) model (e.g. CTM, TTM, and LTM) and enables us to set realistic lengths for
the signal control cycles. Results show, the SCRC model not only attains accuracy comparable to the
CSDT model but also more resilient against extreme traffic conditions. Furthermore, the presented
approach substantially reduces computational complexity and can attain solution faster.