Learning-based traffic signal control algorithms with neighborhood information sharing: An application for sustainable mobility

This research applies R-Markov Average Reward Technique based reinforcement learning (RL) algorithm, namely RMART, for vehicular signal control problem leveraging information sharing among signal controllers in connected vehicle environment. We implemented the algorithm in a network of 18 signalized intersections and compare the performance of RMART with fixed, adaptive, and variants of the RL schemes. Results show significant improvement in system performance for RMART algorithm with information sharing over both traditional fixed signal timing plans and real time adaptive control schemes. The comparison with reinforcement learning algorithms including Q learning and SARSA indicate that RMART performs better at higher congestion levels. Further, a multi-reward structure is proposed that dynamically adjusts the reward function with varying congestion states at the intersection. Finally, the results from test networks show significant reduction in emissions (CO, CO₂, NO_x, VOC, PM₁₀) when RL algorithms are implemented compared to fixed signal timings and adaptive schemes.     

Modeling travel time of car with varying demand on an urban midblock section

The present study analyzes the stochastic nature of travel time distribution under the uncertainty of traffic volume and the proportion of cars in the traffic stream. Stochastic response surface method (SRSM) is adopted for modeling the travel time variation under the influence of traffic composition and traffic volume. This model is applied to an uninterrupted urban arterial corridor of 1.7 km length in New Delhi. Video graphic data were collected for 2 days during morning hours between 8 AM and 12 noon and evening hours of 3–7 PM. License plate matching technique was used for measuring the travel time in the study area. This study focused on travel time variation of cars with varying traffic volume and proportion of car in the traffic stream. Linear regression analysis was carried out initially to know the functional relation and significance relation between the input and output variables, and then SRSM analysis was performed. Artificial neural network (ANN) is also considered to map the relation among travel time, traffic volume and composition of traffic stream. A comparative evaluation is made among ANN, SRSM and regression analysis. Results indicate that apart from traffic volume, the influence of car population is more on travel time variation than motorized two-wheelers. It is attributed to the smaller size and comparability better operating condition of motorized two-wheelers. Also, the ANN and SRSM models are more efficient for analyzing the stochastic relation between the response and uncertain explanatory variable than the regression model.     

Optimization models to characterize the broadcast capacity of vehicular ad hoc networks

Broadcast capacity of the entire network is one of the fundamental properties of vehicular ad hoc networks (VANETs). It measures how efficiently the information can be transmitted in the network and usually it is limited by the interference between the concurrent transmissions in the physical layer of the network. This study defines the broadcast capacity of vehicular ad hoc network as the maximum successful concurrent transmissions. In other words, we measure the maximum number of packets which can be transmitted in a VANET simultaneously, which characterizes how fast a new message such as a traffic incident can be transmitted in a VANET. Integer programming (IP) models are first developed to explore the maximum number of successful receiving nodes as well as the maximum number of transmitting nodes in a VANET. The models embed an traffic flow model in the optimization problem. Since IP model cannot be efficiently solved as the network size increases, this study develops a statistical model to predict the network capacity based on the significant parameters in the transportation and communication networks. MITSIMLab is used to generate the necessary traffic flow data. Response surface method and linear regression technologies are applied to build the statistical models. Thus, this paper brings together an array of tools to solve the broadcast capacity problem in VANETs. The proposed methodology provides an efficient approach to estimate the performance of a VANET in real-time, which will impact the efficacy of travel decision making.