https://orcid.org/0000-0002-7660-2935
![Sample our Engineering & Technology journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5933/TOC-Subject-Sample-2021-Engineering-Tech.jpg"})
Abstract
Information of Connected Vehicles (CVs) could describe vehicular dynamics in much greater detail, enhancing the effectiveness of traffic control systems. One important such system is perimeter control, which can achieve better performance by incorporating the evolution of congestion into the identification of protected regions through a dynamic approach. However, little attention has been given to identifying such dynamic regions by developing CV-based network partitioning models in a spatiotemporal dimension. To address this gap, this paper proposes a three-module framework that (1) collects the relevant information of CVs, (2) performs initial partitioning based on some rational considerations, and (3) identifies the optimal protected regions through a partitioning evaluation, improvement, and iteration algorithm. The carried-out comparisons between perimeter control systems employing the resulting protected regions and those using static regions confirm that the proposed framework enhances the efficiency of perimeter control, even for CVs' penetration rates that are as low as 15%.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 In the context of traffic, time-value refers to the concept that the time spent traveling or waiting in traffic has value for individuals and society. In fact, time spent in traffic can result in various costs, such as total time wasted, increased fuel consumption, and reduced quality of life. Therefore, buses, for example, tend to have a higher time-value compared to passenger vehicles because they carry more passengers and have higher operational costs. Additionally, the time-value of buses can also vary depending on their current condition, such as their level of occupancy or whether they are running on schedule.
2 CV-based Partitioning with respect to traffic Heterogeneity and Dynamicity.
3 It has been previously established that ensuring the uninterrupted movement of special vehicles often results in a rise in the average delay of other vehicles (Xie, Wu, and Yang Citation2022). Accordingly, the mentioned objectives are not necessarily compatible with each other.
4 As a case in point, the maximum number of PRs in our simulations will not exceed three (see Section 4).
5 Note that for the sake of simplicity we work in 2-D space, that is we consider a network without any overpass and underpass crossing.
6 Due to the incorporation of limited pieces of information of CVs, we assume here that the collected data is consistent. However, the issue of data inconsistency in real-world applications can be addressed by using emerging data storing techniques like ontology-based data modeling (Moradi, Sebt, and Shakeri Citation2018).
7 We assume that the expected penetration rate of CVs in each link corresponds to the penetration rate of CVs across the network. In reality, the penetration rate of CVs in each link would be dynamically and stochastically changing. However, imposing this assumption is reasonable, since, even in case of equilibrium, the composition of traffic is uncertain and stochastic by nature. Hence, considering CVs as the only source of real-time data, it is impossible to find the exact CV percentage in each link at each time step.
8 Spielman and Teng (Citation2007) have already confirmed the accuracy of this method to approximate the result of the conductance cut.
9 At first, all intersections are located either within or on the boundary of a PR.
10 Notice here that in such optimizations, there is always a trade-off between accuracy and efficiency. Greedy algorithms are selected here due to their efficiency.
11 The reader can find the applied script at https://github.com/HossseinMoradi/Project15.
12 In the determination of timings in such systems, it is customary to assume a guaranteed minimum green time whose value is approximately equal to the time needed for the crossing of pedestrians. Considering the layout of links in the simulated network and the speed of pedestrians (Muley et al. Citation2018), we use these scaling factors to meet this conventional rule.
13 The reader can find the applied script at https://github.com/HossseinMoradi/Project16.
14 Considering the total links' length of 3600 meters for the two-way 75 meter links in the PR of the benchmark (see Figure ), this coefficient results in an optimal density corresponding to 180 vehicles in the PR (as was also found from Figure ).
15 Recall that the reason behind embedding these parameters is to provide flexibility for dealing with different conditions. The settings of these parameters, therefore, depend on the specific network and traffic characteristics. For example, with a larger network or more lanes in each link, we use higher τ or ϖ (respectively).