information

Review
An Adaptive Traffic Signal Control in a Connected
Vehicle Environment: A Systematic Review
Peng Jing, Hao Huang * and Long Chen
School of Automotive and Traffic Engineering, Jiangsu University, Zhenjiang 212013, China;
jingpeng@ujs.edu.cn (P.J.); chenlong@ujs.edu.cn (L.C.)
* Correspondence: villa.huangh@foxmail.com; Tel.: +86-183-5286-7802

Received: 4 August 2017; Accepted: 22 August 2017; Published: 24 August 2017

Abstract: In the last few years, traffic congestion has become a growing concern due to increasing
vehicle ownerships in urban areas. Intersections are one of the major bottlenecks that contribute to
urban traffic congestion. Traditional traffic signal control systems cannot adjust the timing pattern
depending on road traffic demand. This results in excessive delays for road users. Adaptive traffic
signal control in a connected vehicle environment has shown a powerful ability to effectively alleviate
urban traffic congestions to achieve desirable objectives (e.g., delay minimization). Connected vehicle
technology, as an emerging technology, is a mobile data platform that enables the real-time data
exchange among vehicles and between vehicles and infrastructure. Although several reviews about
traffic signal control or connected vehicles have been written, a systemic review of adaptive traffic
signal control in a connected vehicle environment has not been made. Twenty-six eligible studies
searched from six databases constitute the review. A quality evaluation was established based
on previous research instruments and applied to the current review. The purpose of this paper is
to critically review the existing methods of adaptive traffic signal control in a connected vehicle
environment and to compare the advantages or disadvantages of those methods. Further, a systematic
framework on connected vehicle based adaptive traffic signal control is summarized to support the
future research. Future research is needed to develop more efficient and generic adaptive traffic
signal control methods in a connected vehicle environment.

Keywords: traffic signal optimization; intersection control; adaptive signal control; connected
vehicle; VANET

1. Introduction
Recent years have witnessed a tremendous increase in car ownership in urban areas. This has
result in increasing travel time, traffic congestion, gas emissions, and fuel consumption [1–3]. It is
estimated that delays at traffic signals contribute a 5 to 10 percent of all traffic delays, or 295 million
vehicle-hours of delays, on major roadways alone in the USA [4]. Moreover, the current road and
affiliated infrastructure design and operation in cities are inadequate to meet the rising demands of
the traffic [5]. Traffic signal control systems play an important role in optimizing the flow of traffic and
are the primary means for implementing Smart Roads principles within Network Operating Plans
(NOPs) [6]. In order to improve the efficiency of road use and improve the traffic conditions, it is
essential to optimize the traffic signal control in accordance with traffic demand.
The traditional signal control strategies have gone through three stages: fixed-time, actuated,
and adaptive. Fixed-time signal control utilizes the historical traffic data to determine signal timing.
However, in reality, the traffic demand is unpredictable and fluctuates in time [7,8]. The fixed timing
parameter settings cannot meet the requirement of rapidly changing traffic conditions. Actuated
signal control, which is usually applied for isolated intersection, collects real-time traffic data through
infrastructure-based sensors, e.g., loop detectors, video detectors, infrared, or radar, then cycle length,

Information 2017, 8, 101; doi:10.3390/info8030101 www.mdpi.com/journal/information

Information 2017, 8, 101 2 of 24

phase splits, and even phase sequence react to current traffic demand. However, actuated signal
controllers change these timings based on a set of pre-defined, static parameters such as unit extension
time, minimum, and maximum green time [9,10]. Adaptive traffic signal control strategy, which is
applied for an arterial or road network, employs upstream detector data to estimate incoming traffic
flow and seeks an optimal timing strategy to maximize or minimize an objective function. Several
current adaptive signal control systems include SCATS [11], OPAC [12], SCOOT [13], RHODES [14],
PRODYN [15], and MOTION [16]. However, there are two limitations to actuated and adaptive signal
control. First, those detectors only provide instantaneous vehicle information data when a vehicle is
passing over the detector and cannot measure the vehicle states (such as, position, heading, speed, and
acceleration). Second, the installation and maintenance cost of the fixed sensors is considered high.
If one or more loop detectors are not operating, the performance of the adaptive signal control system
might be notably degraded [17].
As early as 2004, Huang and Miller [18] proposed the conception of a smart intersection making
use of wireless communication. A simple and reliable protocol for electronic traffic signaling systems
was presented to construct a sample application: a red-light alert system. Although the system was
not tested at the field intersection, this work provides the motivation to explore the area of wireless
technologies for adaptive traffic control systems. Connected vehicle (CV), as an emerging technology,
can communicate with each other (V2V) and with the infrastructure (V2I) through dedicated
short-range communications (DSRC). Connected vehicle combines several emerging technological
advances, such as advanced wireless communications, on-board computer processing, advanced
vehicle sensors, GPS navigation, and smart infrastructure to provide a networked environment.
Compared to the traditional detectors, CV technology can provide real-time information (such as,
position, speed, acceleration, and other traffic data) necessary for evaluating traffic conditions on a
road network. Connected vehicle technology has the potential to reduce travel time by 37%, reduce
emissions by 30% and improve safety indicators by 45% [1]. As a component of mobility, intersection
traffic signal control has an important influence on the traffic efficiency. Inspired by such benefits, CV
has been attracting increasing attention in traffic signal control. Implementation of adaptive traffic
signal control in connected vehicle environment has been affected by employing sensors for capturing
traffic information. Communication between vehicle and infrastructure enables the intersection
controller to obtain a much more detailed information of the surrounding vehicle states within the
transmission range. Further, data from connected vehicles provide real-time vehicle location, speed,
acceleration, and other vehicle data. This real-time data is used by the traffic signal controller to
make better timing optimization in controlling the traffic signals. Collecting connected vehicle data
is significantly less expensive to install and maintain a suite of detectors (e.g., loop, radar or video).
If one or more connected vehicles cannot communicate to the infrastructure due to one communication
failure or the other, it will only decrease the market penetration rate on a road network and will not
have a large impact to the total signal control system performance. If the infrastructure is out of order
by chance, the intersection control strategy can restore to the traditional actuated or fixed time signal
control quickly [17].
By taking advantage of connected vehicle technology, adaptive traffic signal control can be divided
into two main parts. The first part is to obtain traffic information at intersections; the second part is
to analyze and evaluate the data acquired from the first part to generate the optimal signal control
strategies. Several studies have been implemented on the applications of CVs technology in adaptive
traffic signal control. Some papers [17,19] concentrated on phase optimization-based methods to
optimize the signal control and some [20–23] employed queue-based methods to model and achieve
the signal control system optimization. Adaptive traffic signal control methods are aimed at either
minimizing the average delay per vehicle or decreasing the queue length of vehicles at intersections.
However, it should be noted that the most of early studies assumed all or a majority of the vehicles
are equipped wireless or connected. Only a few recent works took into consideration the incomplete
vehicle status information or unequipped vehicles. Traffic models [17] and statistical methods [24]
Information 2017, 8, 101 3 of 24

are used to estimate the arrival status of unequipped vehicles. Although there has been certain
positive achievement on models and solutions to research the adaptive traffic signal control in a
connected vehicle environment, there are still many questions to be studied. To summarize, connected
vehicle-based traffic signal control methods have the following limitations in the surveyed literature:
(1) some papers did not estimate the performance of the method under the different penetration
rates of equipped vehicle; (2) few papers estimated the status of unequipped vehicles; (3) some
proposed methods could only be implemented to an isolated intersection, rather than coordinated
intersections. Overall, the purpose of this paper is to review systematically existing adaptive traffic
signal control methods in a connected vehicle environment and to objectively compare the advantages
or disadvantages of those methods. A framework of adaptive traffic signal control in a connected
vehicle environment is summarized based on existing research to support future research. Several
recommendations for future research are provided in the end.
The remainder of this paper is organized as follows. Section 2 provides the methods and a review
of eligible papers. Section 3 presents the systematic review process. Section 4 describes the quality of
the reviewed studies. Limitations and strengths of this paper are proposed in Section 5. The discussions
are presented in Section 6. Section 7 provides the conclusions. Finally, the future work is presented in
Section 8.

2. Methods

2.1. Search Strategy

On the basis of the PRISMA [25] (preferred reporting items for systematic reviews and meta-analyses)
guidelines, six databases were searched in May 2017 for peer-reviewed papers with regard to
adaptive traffic signal control in a connected vehicle environment. These included Web of Science,
ScienceDirect, Academic Search Complete, SpringerLink, IEEE Xplore, and TRID. The first four of
these are comprehensive databases, and the other databases include journals for various disciplines,
such as engineering, mathematics, statistics, computer science, and transportation. At least one term
from each of the three categories of search terms or keywords must be used to contain: (1) signal
control, traffic signal control, adaptive signal control, real-time signal control, intersection control,
traffic light; (2) intersection, isolated intersection, signalized intersection; and (3) connected vehicle,
Vehicular Ad hoc Networks, VANET, Internet of vehicle, cooperative vehicle, vehicle to vehicle
communication, vehicle to infrastructure communication, Intelligent Traffic System, ITS. The item
search forms were adjusted to match the specific structure and requirement of each database. Duplicate
and irrelevant papers were eliminated and reference lists within selected papers were also researched
for further studies.

2.2. Inclusion and Exclusion Criteria

In order to ensure that the papers are in compliance with the subject of the review, all papers must
be screened again carefully. Concrete contents are met as follows: (1) be published in a peer-reviewed
English journal; (2) related references must be included; (3) must be in a connected vehicle environment
(communication between vehicle and vehicle or infrastructure); (4) must be adaptive traffic signal
control (rather than other control method); (5) optimizing the intersection traffic signal control.

2.3. Data Extraction

Standardized data extraction table was extracted from the reviewed papers using matrix method
of literature review. Information drew from each reviewed paper included literature characteristics
(e.g., author, year, country, and journal), methodology (e.g., estimation of unequipped vehicle status,
number of objective function, applied range of the method), simulation platform (e.g., simulation
software, simulation software property), and simulation and simulation result (e.g., simulation
data source, simulation scenario, comparison of simulation results). To ensure the reliability and
Information 2017, 8, 101 4 of 24

validity of data extraction, the author chose 8 papers from the selected papers and extracted the
data independently. Divergence between the authors about the extracted data was discussed until a
consensus was reached. Ultimately, the authors approved of 85% of the extracted data, indicating high
reliability and validity.

2.4. Quality Assessment

The quality of the selected papers in this review is systemically assessed through a modified
checklist. Papers were assessed on 8 different criteria divided over two factors: the methodology
quality and the simulation or simulation result quality.
In order to find out the key factors that influence the quality of methodology, the modified
checklist is divided into three aspects: (1) estimation of unequipped vehicle status; (2) number of
objective function; and (3) applied range of the method. The estimation of unequipped vehicle status
is important to the methodology. This is because the penetration rate of the equipped wireless vehicle
will not reach up to 100% during recent years and the unequipped vehicle will have an influence on
the intersection traffic signal control efficiency [17,26]. Several various objective functions were taken
into consideration when designing a signal control system, such as minimizing the average delay of
vehicle [27], minimizing the delay of total vehicle, minimizing the queue length [17] and improving the
average speed [24]. The alternatives of objective function will determine the performance of the traffic
signal control system. Additionally, in reality, the road intersection is not isolated and the signal control
needs to be realized the coordinated control. Some papers only consider an isolated intersection, so the
applied range of the method is selected to one of the criteria.
To assess the influences affecting the quality of simulation and simulation result, the modified
checklist includes: simulation testing, simulation data source, simulation scenario, penetration rate
of the equipped vehicles, and comparison of simulation results. Since the testing process is not only
increasingly expensive, but also extremely time consuming. Traffic simulation, as an economical, safe,
repeatable, and controllable tool, can model the traffic in a virtual environment to verify the proposed
method. The field data source and the field scenario make the simulation results more authentic
and reliable. Therefore, it is essential to assess the simulation within the selected papers. Moreover,
assessing the performance of traffic signal control system under different penetration rate conforms to
the reality of the development of connected vehicle technology. Comparison of simulation results is to
show whether the proposed method is the best.
All included studies were evaluated on the basis of the 8 criteria listed in Table 1.

Table 1. Checklist for evaluating studies’ quality.

Criteria Description Score

Assessing methodology quality 0–3
Included 1
Estimation of the equipped vehicle status
Not Included 0
More than 2 1
Number of the objective function
Less than 2 0
For an isolated intersection 0
Applied range of the method
For coordinated intersections 1
Assessing simulation and simulation result quality 2–7
Testing 2
Simulation testing
Not testing 1
Field 1
Simulation data source
Hypothetical 0
Field 1
Simulation scenario
Hypothetical 0
100% 1
Penetration rate of CV
Not 100% 2
Included 1
Comparison of simulation results
Not Included 0
Information 2017, 8, 101 5 of 24
Information 2017, 8, 101 5 of 24

3.
3. Systematic
Systematic Review
Review Process
Process
The
The search
search and
and retrieval
retrieval process
process is
is shown
shown inin Figure
Figure 1.
1. The
The number
number ofof papers
papers collected
collected from
from each
each
database
database above were 635 (Web of Science), 1672 (ScienceDirect), 903 (Springer Link), 618Xplore),
above were 635 (Web of Science), 1672 (ScienceDirect), 903 (Springer Link), 618 (IEEE (IEEE
1365 (TRID),
Xplore), and 29 and
1365 (TRID), (Academic SearchSearch
29 (Academic Complete). After
Complete). duplicates
After duplicateswere
wereremoved,
removed,a atotal
total of
of
5223 different records were extracted from six databases, of which 357 were identified
5223 different records were extracted from six databases, of which 357 were identified following the following
the screening
screening of titles
of titles andand abstracts.
abstracts. There
There were
were threereasons
three reasonsfor
foreliminating
eliminatingirrelevant
irrelevant and
and ineligible
ineligible
papers:
papers: not about signal control; not in a connected vehicle environment; and the full text is
not about signal control; not in a connected vehicle environment; and the full text is not
not
available. Thus, the full text of 23 publications was retrieved. The reference lists of excluded
available. Thus, the full text of 23 publications was retrieved. The reference lists of excluded reviews reviews
were
were reviewed
reviewed andand potential
potential papers
papers were
were gathered. Finally, 26
gathered. Finally, 26 published
published papers
papers matching
matching all all the
the
criteria
criteria were
were included
included in in this
this review,
review,asasshown
shownin inTable
Table2.2.

Springer Link IEEE Xplore Academic

Web of n = 903 n = 618 Search
Science Complete
n = 635 ScienceDirect TRID
n = 29
n = 1673 n = 1365

Non-duplicated papers
n = 3869

Not about intersection

Screening titles and abstracts signal control
n = 357 n = 52
Not in a connected vehicle
environment
Reviews excluded Full-text papers for eligibility n = 292
n=2 n = 23 Full-text is not available
n=2

Screening reference lists

n=5

Total papers included

n = 26

Figure 1. The flowchart of systematic review process.

Figure 1. The flowchart of systematic review process.

3.1. Adaptive Signal Control Methods

3.1. Adaptive Signal Control Methods
In this section, we emphasize the adaptive traffic signal control methods implemented in a
In this section, we emphasize the adaptive traffic signal control methods implemented in a
connected vehicle environment. Further, these methods are analyzed and compared objectively. As
connected vehicle environment. Further, these methods are analyzed and compared objectively.
shown in Table 3, these methods can be divided into control methods for isolated intersection and
As shown in Table 3, these methods can be divided into control methods for isolated intersection and
multiple intersections according to the applied scope of the methods.
multiple intersections according to the applied scope of the methods.
Information 2017, 8, 101 6 of 24

Table 2. Summary of studies included in this systematic review.

Penetration Simulation Simulation

Leader Author (Year) Country Journal Method Objective Functions Data Resource
Rate Scenario Platform
Transportation Research Part C: a Distributed-Coordinated 1. the queue length
Islam S.M.A.B.A. 2017 USA - Hypothetical Hypothetical VISSIM
Emerging Technologies methodology 2. the throughput
a reinforcement learning-based
IEEE Transactions on traffic control that integrates V2X 1. the throughput SUMO
Liu W. 2017 China - - Field
Vehicular Technology networks’ dynamic clustering 2. the average waiting time NS-3
algorithm.
IEEE Transactions on a fuzzy group-based intersection
Cheng J. 2017 China the average waiting time - Hypothetical Hypothetical NS-3
Industrial Informatics control
Frontiers of Information an adaptive green traffic signal
1. the vehicle waiting time
Shaghaghi E. 2017 Malaysia Technology & Electronic controlling using vehicular - - Field Veins
2. the pollutant emission
Engineering communications (AGTSC-VC)
an intelligent traffic light
IEEE Transactions on controlling algorithm and an 1. the delay SUMO
Younes M.B. 2016 Canada - Hypothetical Hypothetical
Vehicular Technology arterial traffic light controlling 2. the throughput NS-2
algorithm
a multi-agent based control
Xiang J. 2016 China Cluster Computing the total travel time - Field Field VISSIM
method
Transportation Research Part C: a person-delay-based optimization
Hu J. 2015 USA the delay - Field Field VISSIM-COM
Emerging Technologies method for TSP
Transportation Research Part C: a real-time adaptive phase 1. the total vehicle delay 100%, 75%, 50%,
Feng Y. 2015 USA Field Field VISSIM-COM
Emerging Technologies allocation algorithm 2. the queue length 25%
IEEE Transactions on
a queue length estimation based
Tiaprasert K. 2015 USA Intelligent Transportation the queue length 80%, 50%, 10% - - VISSIM
adaptive signal control
Systems
Transportation Research Part C: an intersection traffic control 1. the total delay
Guler S.I. 2014 Switzerland 100–0% Hypothetical Hypothetical MATLAB
Emerging Technologies algorithm 2. the total number of stop
Transportation Research Part C:
He Q. 2014 USA a multi-modal traffic signal control the delay - Field Field VISSIM-COM
Emerging Technologies
IEEE on Intelligent
an agent-based online adaptive 1. the travel delay
Kari D. 2014 USA Transportation Systems - - - SUMO
signal control (ASC) strategy 2. the fuel consumption
Conference
a cumulative travel-time
Journal of Transportation 1. the total delay time 100%, 90%, 70%,
Lee J. 2013 USA responsive (CTR) field-time Hypothetical Hypothetical VISSIM-COM
Engineering, 2. the average speed 50%, 30%, 10%
intersection control algorithm
Transportation Research the delay or a combination
a predictive microscopic 100%, 50%, 25%,
Goodall N. 2013. USA Record Journal of the of delay, stops and Field Field VISSIM-COM
simulation algorithm (PMSA) 10%
Transportation Research Board decelerations
Information 2017, 8, 101 7 of 24

Table 2. Cont.

Penetration Simulation Simulation

Leader Author (Year) Country Journal Method Objective Functions Data Resource
Rate Scenario Platform
SUMO
IEEE Transactions on Vehicular the oldest arrival first (OAF) the average delay per 100%, 90%, 70%,
Pandit K. 2013 USA - - OMNET++
Technology algorithm vehicle 50%, 30%
(Veins)
Journal of Network & an adaptive traffic signal system 1. the average waiting time
Maslekar N. 2013 France - - Field NCTUns
Computer Applications based on car-to-car communication 2. the queue length
a vehicle-to-infrastructure
IET Intelligent Transport
Cai C. 2013 Australia communication-based adaptive the travel time - Hypothetical Hypothetical Commuter
Systems
control’ (VICAC) method
Transportation Research Part C: a model based on Timed Petri Nets
Ahmane M. 2013 France the queue length - Field Field Video 2
Emerging Technologies with Multipliers
1. the average junction
Republic of a method of queue length
Chang H.J. 2013 Ad Hoc Networks waiting time - Hypothetical Hypothetical GLD
Korea estimation
2. the total queue length
Telecommunication Networks a new Intelligent Road Traffic OPNET
Nafi N.S. 2012 Australia the average waiting time - Hypothetical Hypothetical
and Applications Conference Signaling System (IRTSS) system Modeler
Transportation Research Part C: a unified platoon-based 1. the throughput 100%, 80%, 60%,
He Q. 2012 USA Field Field VISSIM-COM
Emerging Technologies mathematical formulation 2. the average delay 40%, 20%
1. the vehicle delay
12th International Conference a passenger-based traffic signal
Chou L. 2012 Taiwan 2. the stop times - Hypothetical Hypothetical NCTUns
on ITS Telecommunications mechanism
3. the passenger delay
a new adaptive traffic light system
1. the stop times
Tomescu O. 2012 Romania U.P.B. Sci. Bull and a new traffic light green-wave - Hypothetical Hypothetical MATLAB
2. the delay
control algorithm
a new control system for traffic
Ezawa H. 2010 Japan Springer Berlin Heidelberg the average delay time - Hypothetical Field Artisoc
signals by vehicle route sharing
100%, 50%, 33%,
Intelligent Transportation a decentralized adaptive traffic
Priemer C. 2009 Germany the total queue length 25%, 20%, 17%, Field Field AIMSUN NG
Systems signal control algorithm
14%, 12%, 10%.
Ns-2 and
IEEE Vehicular Technology
Gradinescu V. 2007 Romania an adaptive traffic light system the total average delay - Field Field Jist/SWANS
Conference-vtc-spring
VISSIM
Information 2017, 8, 101 8 of 24

Table 3. Summary of methods of adaptive traffic signal control utilized in selected papers.

Traffic Signal Controller Mechanisms

Leader Author (Year) Method/Algorithm Consideration Objective Functions
Traffic Data Gathering Schedule Technique
Minimizing queue length and
A Distributed-Coordinated Distributed coordinated signal
Islam S.M.A.B.A, 2017 2 and 9 intersections Connected vehicle maximizing throughput and reducing
methodology timing
travel time
A reinforcement learning Cooperative RL-based signal Improving traffic throughput and
Liu W. 2017 3 intersections V2X networks
traffic control scheme control algorithm reducing average waiting time
Fuzzy group-based Vehicle grouping based traffic
Cheng J. 2017 Isolated intersection Vehicular Ad Hoc Networks Reducing the waiting time
intersection control control
Density-based and priority-based Decreasing the vehicle waiting time and
Shaghaghi E. 2017 AGTSC-VC 36 intersections Vehicular Ad Hoc Networks
traffic signal timing pollutant emission
ITLC algorithm/ATL An intelligent traffic light Minimizing the delay and improving the
Younes M B. 2016 Isolated intersection Vehicular Ad Hoc Networks
controlling algorithm controlling algorithm throughput
Reducing average travel time, the
Multi-agent based control Vehicle-to-infrastructure Traffic signal co-learning
Xiang J. 2016 22 intersections average delay and the average queue
method communication optimization algorithm
length
A person-delay-based 2 consecutive
Hu J. 2015 Connected vehicle TSPCV-C Reducing bus delay
optimization method for TSP intersections
Minimizing total vehicle delay or queue
Feng Y. 2015 Phase allocation algorithm Isolated intersection Connected vehicle DP based control algorithm
length
Tiaprasert K. 2015 Queue-based ASC Isolated intersection Connected vehicle Queue based traffic control Minimizing the queue length
An intersection traffic control Discharging sequence based Minimizing total delay and total number
Guler S.I. 2014 Isolated intersection Connected vehicle
algorithm traffic control of stops
Multi-modal traffic signal
He Q. 2014 2 intersections Connected vehicle Mixed integer linear program Reducing the delay
control
Reducing the travel delay and the fuel
Kari D. 2014 Agent-based online ASC Isolated intersection Connected vehicle Flexible traffic light state machine
consumption
Travel-time responsive control Improving the total delay time and the
Lee J. 2013 CTR Isolated intersection Connected vehicle
algorithm average speed
Predictive microscopic
Goodall N. 2013. PMSA 4 intersections Connected vehicle Minimizing the delay
simulation algorithm
Information 2017, 8, 101 9 of 24

Table 3. Cont.

Traffic Signal Controller Mechanisms

Leader Author (Year) Method/Algorithm Consideration Objective Functions
Traffic Data Gathering Schedule Technique
Pandit K. 2013 OAF Isolated intersection Vehicular Ad Hoc Networks On-line Scheduling Algorithm Minimizing the delay
Clustering Vehicular Ad Hoc Density-based adaptive traffic Reducing the waiting time and queue
Maslekar N. 2013 CATS 7 intersections
Networks signal timing length
Vehicle-to-infrastructure Travel-time based control
Cai C. 2013 VICAC Isolated intersection Reducing the travel time
Communication algorithm
Ahmane M. 2013 A model based on TPNM Isolated intersection Cooperative vehicle TPNM based traffic control Minimizing the queue length
A method of queue length Reducing average junction waiting time
Chang H.J. 2013 Isolated intersection Vehicular Ad Hoc Networks A dual ring configuration control
estimation and queue length
Nafi N S. 2012 IRTSS Isolated intersection Vehicular Ad Hoc Networks An adaptive IRTSS Reducing the average waiting time
Improving the throughput and reducing
He Q. 2012 PAMSCOD 8 intersections Probe data Mixed integer linear program
the average delay
Vehicle-to-infrastructure Passenger-based adaptive traffic Reducing vehicle delay, stop times and
Chou L. 2012 PATSC Isolated intersection
Communication signal control passenger delay
Tomescu O. 2012 A new adaptive TLC 3 adjacent intersections Vehicular Ad Hoc Networks Traffic signal green-wave strategy Reducing the stop number and delay
Vehicle route sharing based
Ezawa H. 2010 62 intersections Wireless Communication ETC based control algorithm Minimizing the average delay
ASC
Decentralized adaptive traffic Signalized and Vehicle-to-infrastructure DP&CE based optimization
Priemer C. 2009 Reducing the total queue length
signal control unsignalized intersections Communication algorithm
Gradinescu V. 2007 Adaptive Traffic Light Isolated Intersection Vehicular Ad Hoc Networks Modified Webster’s formula Minimizing the average delay
Information 2017, 8, 101 10 of 24

3.1.1. Control Methods for Isolated Intersection

Gradinescu et al. [28] proposed an adaptive traffic light system based on wireless communication
between vehicle and fixed controller nodes. In the system, the timing plan is generated once during
each cycle and establishes a plan for the following cycle based on the measured parameters. The first
step is to calculate the cycle length using Webster’s formula [29] and adjust it to meet objective function,
then the green splits for each phase are allocated to obtain equal degrees of saturation. Two various
scenario simulation results show the total average delay, fuel consumption, and pollutant emissions
were reduced significantly. However, this paper did not consider the performance of method under
various penetration rates.
Feng et al. [17] presented a real-time adaptive signal phase allocation algorithm based on bi-level
optimization using connected vehicle data. Dynamic programming (DP) [30] is applied to optimize
the phase allocation. In the system, signal phase sequences and durations are assigned based on
predicted vehicle arrivals. The algorithm is a bi-level optimization program: at the upper level,
DP contains a forward and a backward recursion. The forward recursion calculates the optimal phase
duration and value of the optimal function. The backward recursion retrieves the optimal signal policy
starting from the final stage working backwards; at the lower level, optimization is formulated in two
objective functions: minimizing the total vehicle delay and minimizing the queue length. To solve
the optimization problem, the combinations of the phase duration and sequences are enumerated to
minimize delay combination. Another research highlight of the paper is the Estimation of Location and
Speed (EVLS) algorithm. An innovative EVLS algorithm is proposed to estimate unequipped vehicle
status. The road section near an intersection is divided into three regions: queuing region, slow-down
region, and free-flow region. Different algorithms are employed to estimate the location and speed of
each unequipped vehicle in each region. The simulation results show that the proposed algorithm
reduces the total delay significantly under high penetration rates and is comparable to actuated control
under low penetration rates.
Compared with the dynamic programming algorithm, the approximate dynamic programming [31]
adopts the combination of off-line and on-line training, which can respond to the change of system
parameters in real time and enhance the robustness of the system. Cai et al. [32] presented a
method consisting of travel-time estimation and adaptive traffic signal control under V2I environment.
The proposed method is based on approximate dynamic programming (ADP), which allows the traffic
controller to learn from its own performance progressively. In the method, the remaining travel time
will be predicted when a vehicle approaches the stop line at an intersection. The approximation
techniques can mitigate the difficulties of solving the dynamic programming to optimize the control
performance. Based on function approximation which is one of the common methods for ADP, they
design a continuous function to replace the exact one. Compared to the benchmarking control method,
the numerical experiment shows that the reduction in the total travel time and vehicle stops is notable
under various simulation scenarios.
Similar to the above method based on travel time, Lee et al. [24] presented a cumulative travel-time
responsive (CTR) real-time intersection control algorithm with connected vehicle data to minimize
the cumulative travel time of vehicles. The cumulative travel time is defined as the elapsed time from
when the vehicle entered the approaching link to the current position of the vehicle. To supplement the
travel-time data obtained at less than 100% market penetration, a stochastic state estimation utilizing
Kalman Filter is used in estimating cumulative travel time. The algorithm calculates the travel time for
the combination of vehicle movements for each phase (i.e., NEMA phases 2 & 6, or 4 & 8). The phasing
with the highest combined travel time is selected as the next green phase, and a minimum green time
is five seconds. The algorithm is tested on an isolated intersection model, and shows improvement of
34%, 36%, and 4% in total travel time, average speed, and throughput at the 100% market penetration
compared to an actuated system. Futhermore, results from the simulation highlight that at least 30%
market penetration rate is critical for the proposed method.
Information 2017, 8, 101 11 of 24

Chang et al. [33] proposed a real-time traffic control system based on VANETs. In the system,
VANETs was used to gather traffic information, then the algorithm estimated the queue lengths,
assigned vehicles to each lane, and optimized cycle lengths and green splits for a traffic signal
controller based on gathered information. To optimize the signal control, a dual ring configuration
is used for adequate phase control and a green time estimation algorithm based on vehicle queue
length is proposed to achieve the signal timing optimization. The simulation results show that the total
waiting queue length is shortened and the waiting time is minimized compared to random control.
Pandit et al. [27] developed the oldest arrival first algorithm (OAF) to minimize the delay across
the intersection by scheduling the optimal sequence of conflicted phases at each traffic light. This paper
formulates the traffic signal as a job scheduling problem, where each job corresponds to a platoon of
vehicles. First, a conflict graph is constructed indicating all competing traffic flows at each isolated
intersection. Then, the rule of “first come first serve” schedules the competing platoons of traffic in
each flow, using the estimated arrival time of each predictable platoon. Mathematical analysis and
simulation implementation indicate that the correctness and benefits of the proposed algorithm over
pre-timed and actuated scheduling traffic signal control method.
Younes et al. [34] proposed an intelligent traffic light controlling (ITLC) algorithm based on
VANETs. In addition to the ITLC algorithm for the isolated intersection, this paper also presented an
arterial traffic light (ATL) controlling algorithm for the arterial road. In the algorithms, vehicular ad
hoc networks technology is utilized to gather the real-time traffic information at each signalized road
intersection. Both ITLC and ATL algorithm optimize the sequence phases and the time of each phase
according to the real-time traffic characteristics of all traffic flows. The experimental results show that
ITLC can decrease the delay by 25% and increase the throughput of each road intersection by 30%.
Moreover, ATL can increase the traffic fluency and the throughput by 70% on the arterial street.
Nafi et al. [35] presented a unique VANET based Intelligent Road Traffic Signaling System (IRTSS)
system which can collect traffic information from individual cars and exchange road traffic information
to dynamically control the traffic signaling cycle. Compared with the previous works, the proposed
IRTSS can optimize the fuel consumption and emission by improving traffic flows. A new traffic
estimation technique has been developed to implement an adaptive signal control method based on the
vehicles density at the intersections. The simulation based on OPNET shows that the proposed method
can achieve significant improvements in waiting time, compared to the fixed-time signal control.
Unlike the optimal signal light control, Tiaprasert et al. [22] presented queue-based adaptive
signal control using connected vehicle technology. In this model, connected vehicle technology is
used to estimate the queue length for adaptive signal control. To estimate the queue, a discrete
wavelet transform is introduced to enhance the consistency of queue estimation for the first time.
The proposed method can be implemented without the assumption of pre-timed signal, signal interval,
and specific arrival distribution. In addition, the volume, queue characteristic, and signal timing
are also not required in the model as basic input data. It is noted that the proposed algorithm is
capable of estimating queue length under various pre-timed signal and actuated signal control for
under-saturated/saturated conditions. The simulation result shows that the proposed queue based
model performs well in the simulation.
Cheng et al. [20] developed a fuzzy group-based intersection control method based on VANETs.
In the method, vehicles in the same lane are divided into small groups and vehicle groups are
scheduled through wireless communication, rather than traffic signal lights. This method has two
advantages against existing algorithms: (1) group-based scheduling reduces average waiting time; and
(2) group-based scheduling improves the grouping fairness. Furthermore, the reinforcement learning
is utilized to adjust the parameters of the network and make it adaptive to various traffic conditions.
The results show that the proposed method can reduce waiting time and improve fairness in various
cases and the advantage against traffic light algorithms can be up to 40%.
Ahmane et al. [21] presented a new traffic control method based on Timed Petri Nets with
Multipliers (TPNM) for an isolated intersection. In this method, the control aims to smooth the traffic
Information 2017, 8, 101 12 of 24

through the sequence of vehicles authorized to cross the intersection. The vehicles arriving at the
intersection receive the traffic “right of way” state information by on-board equipment, the driver then
decides whether to go through the intersection. The proposed method has an excellent performance in
both a real intersection test and a simulation test.
Guler et al. [23] proposed a traffic signal control algorithm utilizing the information from
connected vehicle technology for a single intersection. The proposed algorithm optimizes sequences
of cars discharging from the intersection by incorporating information from equipped vehicles to
minimize the total delay. This paper also studies the effect of automated vehicles by allowing for
priority to switch between approaches rapidly and found only small decrease of delay for low demand
scenarios. The value of platooning and signal adaptability to demand are also evaluated in this paper.
The simulation based on MATLAB shows that the average delay can be reduced up to 60%, with the
penetration rates increasing up to 60%. The result also indicates that the proposed method can also
minimize the total number of stops.
Furthermore, an agent has been widely applied to the various intelligent research fields [36].
The agent-based model has also been introduced to optimize traffic signal control in a connected
vehicle environment because of its utility in studying several agents specified at various scenarios,
capturing the complex dynamic relationships and accounting for feedbacks between individuals their
environments. Kari et al. [37] proposed an agent-based online adaptive traffic signal control (ATSC)
based on connected vehicle technology. In the system, traffic at an intersection is considered to be a
multi-agent system: (1) Vehicle Agent (VA); and (2) Intersection Management Agent (IMA). The VA
is responsible for communicating real-time vehicle information to IMA and the IMA undertakes
communicating with VA within a communication radius, and determining the optimal signal timing.
Based on the user-defined Measure of Effectiveness, a VA need to predict certain information
(e.g., Time-Of-Arrival) in order to provide the IMA with input. The other innovation is to improve the
fixed sequence of traffic signal controller NEMA. A more advanced and flexible finite state machine
is proposed in the paper. The total of 49 states, including main street and side streets, allow for a
variety of signal strategies to implemented by the IMA. The queue length optimizer is presented
to maximize the number of vehicles within a green light. Under both constant and varied demand
scenarios, the system exhibits significant saving in reducing travel time and system-wide fuel economy.
While the majority of above papers optimize traffic performance, few studies take into
consideration passengers’ feelings. Chou et al. [38] proposed a passenger-based adaptive traffic
signal control mechanism. In the mechanism, Road Side Unit is considered as a traffic signal control
agent and vehicular messages including passenger loading information, fuel pollutant emission, and
fuel consumption. The expected arrival time of each vehicle to the intersection is calculated to compare
with remaining green time. The green time will be adjusted according to the above parameters
dynamically. The simulation results show that the proposed PATSC mechanism can improve the
transportation efficiency up to 23.09%, and reduce pollutant emission up to 10.66%.

3.1.2. Control Strategies for Multiple Intersections

Priemer et al. [39] developed a decentralized adaptive traffic signal control algorithm with V2I
communication data. The signal control algorithm is phase-based, but operated without common
parameters like cycle times and offsets. The algorithm seeks to minimize the total queue length
by optimizing phases in five-second intervals over a 20 s horizon using the techniques of dynamic
programming and complete enumeration. To reduce the quality loss due to low or mid penetration
rate, a method for queue length estimation (ql-estimation) [40] is incorporated to the control algorithm
and queue length is estimated at the end of each interval. There are two innovative aspects to this
paper: the first is analyzing various penetration rates; the second is assigning the priority to special
vehicles like public transit or emergency vehicle, the coordinated control with neighbored intersections.
Compared to TRANSYT-7F [41], a reduction of up to 24% in delay and increase by 5% in mean speed
Information 2017, 8, 101 13 of 24

are reported. The simulation results also show that DP&CE with ql-estimation will improve the
performance of method significantly when the penetration rate is below 33%.
Maslekar et al. [7] presented an adaptive traffic signal control system which utilizes V2V
communications in VANETs. In this paper, clustering algorithm is implemented to estimate the
number of vehicles approaching an intersection. Based on the estimated density, an adaptive cycle
time and the green time for different phased at each intersection are calculated. A modifying Webster’s
model is used to generate the cycle time of each traffic light in an adaptive system. In the system,
cycle times and other safety parameters (green, red, and inter-green interval) computation together
constitute car-car communication based adaptive traffic signal which obtain density information from
VANET. The simulation results show CATS has an improvement in terms of average waiting time and
the percentage of vehicles stopping at each road intersection.
Goodall et al. [42] proposed a predictive microscopic simulation algorithm (PMSA) for traffic
signal control. The algorithm received data from connected vehicles including positions, headings, and
speeds, and imported them to a microscopic simulation model to predict the future traffic conditions
in a connected vehicle environment. A rolling horizon strategy of 15 s was chosen to optimize either
delay only or a combination of delay, stops, and decelerations. In the algorithm, PMSA employs
microscopic traffic simulator to simulate vehicles and calculates the objective function delay directly
from vehicle’s simulated behavior. An important characteristic of the proposed algorithm is that it just
requires instantaneous vehicle data and does not reidentify or track vehicles. The simulation results
show the algorithm has much greater improvement at low and midlevel traffic volumes, and performs
worsened under saturated and oversaturated conditions.
Similarly, Shaghaghi et al. [43] presented a new VANET-based adaptive green traffic signal control
system (AGTSC-VC). In the system, signal control is defined into two main steps: (1) VANET-assisted
traffic information gathering; and (2) traffic density assessment and traffic signal timing generation.
The cluster algorithm is employed to calculate the density of vehicle. Three types of packets including
header-packet, reply-packet, and traffic-load-packet are used to provide traffic density information at
the intersection. Density-based and priority-based traffic signal timing method make the performance
of the proposed approach better than the traditional method. The simulation results demonstrate the
superiority of AGTSC-VC in improving the accuracy of vehicle density estimation, decreasing the
waiting delays of vehicles, conspicuously reducing the gas emission rates, and decreasing the travel
time of prioritized vehicles.
Islam et al. [44] proposed a Distributed-Coordinated methodology for signal timing optimization
in a connected vehicle environment. In the method, the signal timing optimization problem is
reformulated based on a central architecture, where all signal timing parameters were optimized
in one mathematical program. As a result of this distribution, a mathematical program only controls
the timing of a single intersection. Based on this method, the complexity of the traffic signal control
problem was significantly reduced. The simulation demonstrates that the proposed algorithm can
increase intersection throughput between 1% and 5%, and decrease travel time between 17% and 48%,
compared to actuated coordinated signals.
Liu et al. [45] presented a reinforcement learning traffic control method integrating a dynamic
clustering algorithm. In this paper, a dynamic clustering algorithm is proposed to achieve a relatively
stable cluster structure and enhance traffic data collection efficiency. By integrating the clustering
algorithm, a cooperative reinforcement learning control scheme is utilized to optimize the road traffic
and signal control. The simulation based on SUMO shows that the proposed method can effectively
improve the throughput, reduce the average waiting time and avoid traffic congestion compared to
the traditional adaptive signal control method.
Compared to the single traffic modal considered in the above papers, He et al. [46] presented the
platoon-based arterial multi-modal (transit and passenger car) signal control method called PAMSCOD
under V2I environment. In the proposed method, a headway based platoon recognition algorithm is
developed to identify pseudo-platoons based on the online probe data. A mixed integer linear program
Information 2017, 8, 101 14 of 24

problem was used to optimize phasing sequence and start time of phases for the next considered cycle
based on the current traffic controller status, traffic conditions, platoon data, and priority requests.
The simulation based on VISSIM shows that the proposed PAMSCOD can reduce the vehicle delay for
both under-saturated and oversaturated traffic condition. In particular, the result also indicates that a
40% penetration rate is critical for the performance of state-of-practice signal control methods.
To address the conflicting problems in the above PAMSCOD due to different control objectives,
He et al. [47] integrated multi-modal priority control method including emergency vehicles, transit
buses, commercial trucks, and pedestrians, with the consideration of actuated-coordination. In the
proposed method, a request-based mixed-integer linear program (MILP) formulation is utilized
to accommodate multiple priority requests from different modes and optimize the signal timing.
Further, the signal coordination is achieved based on integrating virtual coordination requests for
priority in the formulation. However, it should be noted that the communication between different
traffic modes, especially pedestrian-vehicle/infrastructure communication, was assumed in this paper.
The simulation results demonstrate that the proposed method can reduce bus delay by 24.9% and
pedestrian delay by 14%.
Hu et al. [48] proposed a person-delay-based optimization method for transit signal priority (TSP)
which enables bus/signal cooperation and coordination among consecutive signalized intersections
under the connected vehicle environment. A Binary Mixed Integer Linear Program is used to solve the
bus delay from an upstream intersection to downstream intersections, and thus minimizes personal
delay for all users. In the methods, coordinated TSP with connected vehicle (TSPCV-C) logic is
designed to achieve transit-signal cooperation and coordination among intersections. Simulation
results show that the proposed TSPCV-C logic reduces the bus delay between 55% and 75%, compared
to the conventional TSP. However, the system may not be used for cases with multiple conflicting bus
lines and multiple priority requests under its current form.
The concept of agent has also been used to optimize multi-intersection signal control.
Ezawa et al. [49] proposed an adaptive traffic signal control based on vehicle route sharing. The vehicle
route sharing was to share position and path information, and the route sharing information was used
for calculating expected traffic congestion. In the system, each traffic signal control agent has two
traffic controllers: Cycle-Split controller (CS-controller) and Offset control (O-controller). The former
optimizes traffic signal cycle length and split based on calculating the average and total expected traffic
congestion respectively. The O-controller will activate the offset cooperation if the ratio of expected
traffic congestion of an inflow link to outflow links. The simulation results show that the proposed
methods outperform other traditional traffic signal control strategies.
Xiang et al. [50] presented a novel multi-agent based control method for an integrated network of
adaptive traffic signal controllers under V2I communication environment. There are two innovations
in the system: (1) the novel grid and mixed truncated Gauss model is suitable for parallel processing;
and (2) co-learning provides the recommended shortest time path. The intersection is treated as an
agent and a Markov decision process is used for modelling the intersection of an agent with its own
environment. Further, the agents interact with the environment by trying out actions and use resulting
feedback to reinforce behavior that leads to a desired outcome. The traffic signal control is based on the
following parameters to realize the optimization: vehicle state, action, objective function, and iterative
update rules. The simulation results show that the average travel time per vehicle, the average delay
per vehicle and the average queue length reduce significantly compared to the traditional traffic signal
control method.
Compared to the above papers, Tomescu et al. [51] took into consideration the driver behavior
and new parameters including weather, vehicle type, and road event, and proposed a new adaptive
traffic light system and a new traffic light green-wave control algorithm. In the system, Webster’s
equation [29] is used to calculate the cycle length for each intersection and the maximum cycle length
is selected as the cycle length of entire system. In the determination of optimal offset, the fuzzy logic
algorithm is used to adjust the offset based on weather, vehicle type, and road events, owing to the
Information 2017, 8, 101 15 of 24

algorithm better capturing human expertise. The evaluation shows that the proposed method can
significantly reduce stop number and each car’s delay.
Although scholars utilized various models and algorithms to study adaptive traffic signal control
in a connected vehicle environment from different perspectives, the ultimate goal was to reduce the
delay and improve the overall traffic efficiency by optimizing the road traffic signal control.

3.2. Estimation of Unequipped Vehicle Status

Even though some connected vehicle-based adaptive traffic signal control methods perform better
at higher connected vehicle penetration rates, many are unable to outperform traditional applications
at low penetration rates. The penetration rate of equipped vehicles has a great influence on spatial
temporal dispersion of equipped vehicles [52], so the penetration rate may be a critical parameter in
determining the effectiveness of signal control algorithms. Therefore, it is necessary to estimate the
unequipped vehicle status to improve the performance of the method.
Priemer et al. [39] integrated queue length estimation (ql-estimation) [40] into the proposed
control algorithm. Two different scenarios are considered in the system. In a high penetration rate,
the queue length is directly calculated. However, in a low or mid penetration rate, queue length is
estimated based on the data from multiple sensors technology. Both scenarios work at the end of
each interval. The objective of the signal control is to determine the optimal phase sequence in order
to minimize the total queue. The simulation shows that Dynamic Programming (DP) & Complete
Enumeration (CE) with ql-estimation improves DP&CE’s performance significantly with a penetration
rate below 33%. The performance of DP&CE with ql-estimation is nearly equal to the TRANSYT-7F at
a penetration rate of 20%.
Lee et al. [24] applied the Kalman filtering (KF) technique to estimate traffic states for cumulative
travel-time under imperfect connected vehicle penetration rates. Owing to the result of the feedback
characteristic of the KF technique to recursively correct the errors within the boundary set by both
process and measurements noise variances. The KF technique outperformed various similar methods
in dealing with either traffic status estimation. In this paper, the KF algorithm is comprised of
two equations: (1) a state-space equation; and (2) a measurement equation. The former affirms that
the current states resulting from the previous states, that is, the previous input actions and noises
that occurred at the previous time period. The measurement equation explains that the current
measurements can be obtained from the current estimated states, or vice versa with some noise.
To construct a complete prediction arrival table, the location and speed of each vehicle including
unequipped vehicles on the roadway needs to be estimated. Feng et al. [17] proposed an algorithm
called EVLS to estimate vehicle status of unequipped vehicles. In the system, the road segment near
an intersection is divided into three regions: queuing region, slow-down region, and free-flow region.
The location and speed of each unequipped vehicle in each region are estimated based on different
algorithms. In a queuing region, the locations and stopping times of the last stopped vehicle and
the second to the last stopped connected vehicle are used to estimate the unequipped vehicle. In the
slow-down region, Wiedemann’s car following model is selected to estimate the state of unequipped
vehicles which react with their leading vehicle rationally. In the free flow region, the vehicles are
assumed to behave independently and not interact with other vehicles. The number of equipped
vehicles is divided by the penetration rate to obtain the total number of vehicles. The speeds of the
unequipped vehicles are assumed to be either the posted speed limit or the average speed of the
connected vehicles observed in the field.
Three methods estimate the status of unequipped vehicles from different aspects: queue,
travel-time, location, and speed. Further, those methods play a positive role in studying the
performance of traffic signal control under different penetration rate and the impact of unequipped
vehicles on traffic signal control in a connected vehicle environment.
Information 2017, 8, 101 16 of 24

3.3. Simulation Platform

The field traffic validation is so complex and the testing process is not only increasingly expensive,
but is also extremely time consuming. Consequently, time and site condition restrictions often
render field tests cumbersome, which is why simulations are an indispensable part of the overall
research. In order to demonstrate the effectiveness of the proposed method, different simulation
platforms are employed to simulate various testing scenario. In a connected vehicle environment,
the selected simulation platform needs to meet two basic conditions: (1) an event-based network
simulator achieves vehicle-to-vehicle and vehicle-to-infrastructure communication; and (2) a road
traffic simulator simulates road condition and traffic demand. A summary of simulation platform and
basic description is shown in Table 4.

Table 4. Summary of simulation platform and basic description in the selected papers.

Simulation
Country Introduction Property Frequency
Platform
Veins is a framework for running vehicular network
Open
Veins [53] Germany simulations. It is based on two well-established 2
source
simulators: OMNeT++ and SUMO.
VISSIM is a microscopic, time step and behavior based
simulation model developed to model urban traffic and
VISSIM [54] Germany public transit operations. It is possible to automate Commercial 9
certain tasks in VISSIM by executing COM (component
object model) commands from an external program.
“Simulation of Urban MObility” (SUMO) is a
Open
SUMO [55] Germany microscopic and continuous road traffic simulation 3
source
package designed to handle large road networks.
NCTUns is a network simulation and traffic simulation
software, NCTUns has more realistic and credible
Open
NCTUns [56] Taiwan experimental results, and can directly use the existing 2
source
network software to reduce the workload of the design
of the experimental environment.
Aimsun is traffic modelling software that allows you to
model anything from a single bus lane to an entire
AIMSUN
Spain region. It is used to improve road infrastructure, reduce Commercial 1
NG [57]
emissions, cut congestion and design urban
environments for vehicles and pedestrians.
Green Light District (GLD) simulator is a Java-based
GLD Open
Netherlands traffic simulator that enables road/intersection design 1
simulator [58] source
and allows the expansion of source codes.
MATLAB is a high-level technical computing language
and interactive environment for algorithm development,
MATLAB [59] USA Commercial 2
data visualization, data analysis and numerical
computation, including MATLAB and Simulink.

In this section, we will focus on several traffic simulation platforms which are widely used in a
connected vehicle environment. VISSIM [54], with its powerful and mature simulation module, is the
preferred platform for researchers to simulate intersection traffic signal control scenarios. The COM
(component object model) and API (Application Programming Interface) interface provide powerful
secondary development capabilities to implement vehicle-vehicle or infrastructure communication
under a connected vehicle environment. In [17,22,24,42,44,46–48,50], VISSIM is used to simulate the
intersection traffic signal control scenarios based on the proposed methods. VISSIM itself can model
roads, intersections, vehicle characteristics, car following models, etc. To achieve the communication
between vehicle and vehicle/infrastructure, drivermodel.dll API and virtual ASC controller are added
in [17]. In [42], a program based on C# programming language is added through VISSIM-COM to
extract individual vehicle characteristics. In [24], to incorporate the simulation of connected vehicle
environment and the implementation of the Kalman filter algorithm, MATLAB is employed to connect
with VISSIM through COM interface. VISSIM-COM, the ASC virtual controller as the simulation
Information 2017, 8, 101 17 of 24

environment, and GAMS/CPLEX as optimization solver, constitute the entire simulation platform
in [46,47]. In the simulation process, COM starts a VISSIM simulation and sends the data to CPLEX.
After retrieving optimal plan, COM implements the signal timing by sending phase control commands
to ASC controller. In [22,50], VISSIM is selected as a simulation platform to generate individual
vehicle’s information and estimate the average travel time or queue length as an output.
Compared to the above software, SUMO [55], as an open source tool, provides a wide range of
traffic planning and simulation functionalities to support scientific research. In [37], SUMO is used
together with the Comprehensive Modal Emissions Model (CMEM) through the python Traffic Control
Interface (TraCI). Furthermore, SUMO can couple other network simulation software to achieve V2X
environment simulation platform. In [45], a network simulator NS3 and SUMO are connected by the
TraCI interface. During the simulation, the VANET nodes in NS3 is used to execute the data and the
fixed nodes corresponding to intersections generate the traffic data information and produce adaptive
traffic signal control commands. The data and control commands will be sent back to the traffic
simulator SUMO through TRACI. Moreover, SUMO couples network simulation software OMNeT++
into the V2X simulation platform Veins [53] through the TRACI interface. In [27,43], SUMO is used
to model the road and traffic characteristics. The TraCI interface is utilized for synchronizing the
generated traffic scenarios of SUMO into the OMNET++ simulator to implement V2X.
NCTUns [56], with its high fidelity and well scalability, becomes many research scholars’ choice.
As compared with VISSIM and SUMO, NCTUns provides users with a simulation environment in
which traffic simulation and network simulation are tightly integrated. In [7,38], NCTUns is used to
model the proposed traffic signal control methods to gather the traffic data information and, at the
same time, estimate the related parameters, such as the density, fuel consumption, etc. However, it has
been commercialized as a simulation software EstiNet.
AIMSUN NG [57] is the first to integrate macro, meso, and micro models into a single software.
It can provide a convenient secondary development platform to complete the complex simulation
tasks under a connected vehicle environment. In [39], the proposed traffic signal control algorithm
based on C++ programming language is implemented in AIMSUN NG simulator through AIMSUN
NG API. Via the API module, AIMSUN NG provides the algorithm module with current traffic data,
and in turn, the optimal results are sent to the microscopic simulator.
The Green Light District (GLD) simulator [58], based on Java, allows users to add infrastructures,
to set different traffic patterns, and to evaluate the controllers using different statistical measures
(such as average waiting time). In [33], to achieve the inter-vehicle communication, a packet-based
communication simulator is added to GLD simulator. New algorithms for traffic signal control are
added to GLD simulator through the expansion of source codes.
Moreover, MATLAB [59] is a powerful mathematical modeling software that simulates the various
complex models. MATLAB, in general, is used as a joint simulation with traffic simulation software
based on certain API. In [51], to calculate the offset adjustment constant, fuzzy logic program is coded
in MATLAB to evaluate and the optimal offset parameter. MATLAB is utilized to model and estimate
the proposed intersection control algorithm in [23]. Compared to the aforementioned professional
traffic simulation software, MATLAB cannot show relatively realistic traffic simulation scenarios.

4. Quality and Reviewed Studies

The scores of the quality of each eligible paper range from 3 to 11, as shown in Table 5. According
to Table 5, only 3 studies (11.54%) estimated the unequipped vehicle status in an imperfect connected
vehicle penetration rate. In the selected papers, 14 studies (53.85%) employed more than two objective
functions to optimize the proposed traffic signal control method. All studies have been simulated to
verify the reliability of the proposed method. However, in all selected papers, the simulations were not
all based on the field data and field scenarios. 9 studies (45.00%) and 13 studies (56.52%), respectively,
used the field data and field scenarios to simulate the adaptive traffic signal control method. In view of
penetration rate, 8 studies (30.77%) took into account the performance of the proposed method under
Information 2017, 8, 101 18 of 24

different penetration rates. Nearly all studies (n = 25, 96.15%) compared results with other traffic signal
control methods in order to highlight the superiority of the proposed method.

Table 5. Distribution of quality characteristics across reviewed studies.

Criteria Description Score N of Studies Percentage

Assessing methodology quality
Estimation of the equipped Included 1 3 11.54%
vehicle status Not Included 0 23 88.46%
More than 2 1 14 53.85%
Number of the objective function
Less than 2 0 12 46.15%
For coordinated intersection 1 12 46.15%
Applied range of the method
For an isolated intersection 0 14 53.85%
Assessing simulation and simulation result quality
Testing 2 26 100.00%
Simulation testing
Not testing 1 0 0.00%
Field 1 9 45.00%
Simulation data source
Hypothetical 0 11 55.00%
Field 1 13 56.52%
Simulation scenario
Hypothetical 0 10 43.47%
Not 100% 2 8 30.77%
Penetration rate of CV
100% 1 18 69.23%
Included 1 25 96.15%
Comparison of simulation results
Not Included 0 1 3.85%

5. Limitations and Strengths

The following limitations should be considered when interpreting the existing results. First, we
limited our search to papers published in English, thus, relevant literature published in other languages
was excluded. Second, all included studies were adaptive traffic signal control, while other traffic
signal control methods applied in a connected vehicle environment were not be discussed. This study
had several strengths. First, it used an extensive search strategy to locate papers in six databases and
rigorously screened papers through well-defined inclusion/exclusion criteria. Second, the quality of
included papers was assessed in a standardized way.

6. Discussion
Adaptive traffic signal control based on a connected vehicle environment is a relatively emerging
research field. Compared with traditional traffic signal control methods, this method has many
advantages on reducing the delay and improving the road traffic flow efficiency. From the abovementioned
review, the essence of optimizing the traffic signal is to minimize the vehicle delay at the intersection
by retiming traffic signals or optimizing vehicle queue. Minimizing the delay can reduce the waiting
time for vehicles, smooth the traffic flow at intersections, and reduce the exhaust emissions. In brief,
this method not only improves the efficiency of the road transport system, but also reduces the fuel
consumption and gas emissions.
An adaptive traffic signal control framework is summarized based on the existing research,
with aims to support future research, as shown in Figure 2. The framework is divided into three
modules: input, optimization, and output. In the input module, basic information including vehicle
information, intersection information, proposed intersection signal control method, and objective
functions will provide a basic input setting for optimization module. It is to be noted that the
car-following model in a connected vehicle environment may be different from traditional models.
In the optimization module, the simulation software will calculate the related parameters, estimate
Information 2017, 8, 101 19 of 24

unequipped vehicle status, and determine the arrival time of each vehicle. Based on the above process,
the adaptive traffic signal control will be optimized based on the objective functions. In the output
module, the optimum parameter will be sent to the optimization module, including optimum trajectory
of each vehicle and other pre-setting output parameters. Moreover, the module provides the results of
a given objective function for the authors to evaluate the proposed methods.
Information 2017, 8, x 19 of 24

Input Optimization Output

The estimated values:

Basic information: The calculation of related travel time, density, queue
Vehicle physical parameters: density, travel length
characteristic, time, etc.
Car-following model,
Equipped or not,
Road lane setting condition,
Speed limit
Optimum vehicle trajectory
(the speed ,acceleration,
The prediction of arrival and headway)
time to the intersection and
unequipped vehicle status
Real-time data from
connected vehicle
technology
Optimal intersection signal
control parameters: cycle,
phase sequence, or queue
The optimization of sequence, queue length
Proposed methods: intersection signal control:
intersection traffic control signal timing or queue
strategies and optimization
objective functions

The results of given

objective function: the
delay, the stop and so on

Figure 2. The framework of adaptive traffic signal control in a connected vehicle environment.
Figure 2. The framework of adaptive traffic signal control in a connected vehicle environment.

Although there has been a fruitful development of models and solution techniques to research
Although
the adaptive theresignal
traffic has been a fruitful
control development
in a connected of models
vehicle and solution
environment, techniques
there are still many to research
research
the adaptive
questions, traffic
such signal
as the control in a connected vehicle environment, there are still many research
following:
questions, such as the following:
 The existing researches were mainly focused on the optimization-based method. This literature
• The existing
optimized researches
the werecontrol
traffic signal mainlythrough
focusedbuilding
on the optimization-based
the optimization model method. This literature
of traffic dynamic
optimized
and traffic the traffic
signal signal
control control There
strategy. through arebuilding the optimization
still opportunities model of traffic
for optimization dynamic
problems that
and traffic signal
are solved. controlthe
For instance, strategy.
machine There are still
learning opportunities
method for optimization
has been widely problems
used in various fieldsthat
of
are solved. For instance, the machine learning method has been widely used
artificial intelligence. The method based on machine learning can optimize the controller’s policy in various fields of
artificial
through intelligence.
trial-and-error Theinteractions
method based with onanmachine learningwhich
environment, can optimize the controller’s
is in accordance with policy
traffic
through trial-and-error
signal strategy: optimizinginteractions
the signalwith an environment,
timing through thewhich trafficisflow
in accordance
dynamic. with
However,traffic only
signala
strategy:
few authors optimizing the signal
have employed timing
this method through the traffic
to optimize the flow
trafficdynamic. However,
signal control in a only a few
connected
authors
vehicle. have employed this method to optimize the traffic signal control in a connected vehicle.
• The proposed
The proposed models
models usedused to optimize
to optimize thesignal
the traffic traffic signal
control mightcontrol mightand
be complex, be computationally
complex, and
computationally
expensive, expensive,
sensitive sensitive
to modelling to modelling
errors. errors. Itto
It is necessary is necessary to highly
highly improve theimprove
robustness, the
robustness,and
versatility, versatility,
precisionand precision
of the proposed of the proposed
model. model. Furthermore,
Furthermore, in order
in order to facilitate thetovalidation
facilitate
thethe
of validation
models, of the models,
many scholars many
employedscholars employed
a simplified a simplified
road road or
or intersection intersection
model model
for simulation.
for simulation.
However, However,
road and road and
intersections are intersections
relatively more arecomplex
relativelyin more
reality,complex in reality,
which poses which
a challenge
poses
to the aadaptability
challenge toofthe theadaptability
models. of the models.
• From the review,
review, thethe market
marketpenetration
penetrationrate rateofofconnected
connectedvehicles
vehicles waswasa critical
a critical parameter
parameter in
determining
in determining the the
effectiveness
effectivenessof the connected
of the connected vehicle related
vehicle signal
related control
signal algorithms.
control algorithms.The
source [46] showed that a 40% penetration rate is critical for the performance of proposed signal
control method. Goodall [42] also indicated that the required minimum penetration rate for
traffic signal control in a connected vehicle environment was 20–30%. It is necessary to verify
the performance and adaptability of the proposed model under different penetration rates.
 In addition, although [17,24,39] proposed three various methods to estimate the status of
Information 2017, 8, 101 20 of 24

The source [46] showed that a 40% penetration rate is critical for the performance of proposed
signal control method. Goodall [42] also indicated that the required minimum penetration rate
for traffic signal control in a connected vehicle environment was 20–30%. It is necessary to verify
the performance and adaptability of the proposed model under different penetration rates.
• In addition, although [17,24,39] proposed three various methods to estimate the status of
unequipped vehicles from different aspects (queue, travel-time, location, and speed), the research
on the status estimation of unequipped vehicle in a connected vehicle environment was limited.
For the complexity of microscopic driving behavior model, more focus should be on the
unequipped vehicle status based on the car following model.
• In terms of the applied scope of signal control methods, many proposed models can only be
implemented to an isolated intersection, but it cannot achieve the coordinated control of multiple
intersections or the arterial green wave control. However, the intersection is not isolated in
the road network and the adjacent or multiple intersections should achieve synchronization
or coordination control and, ultimately, obtain the global optimal control in the actual traffic
management and control. This problem leads to an important effect on fluency for vehicles on the
road network.
• The existing methods generally consider a single-modal traffic, which ignore integrating
multi-modal traffic or priority for special modes, such as transit, truck, and pedestrians into
the methods. This is one of the future research directions because the traffic flow system is a
human-joined, changeable, and complex system. Therefore, more road traffic factors should be
taken into account in the modeling process.
• The inter-vehicle or vehicle-infrastructure communication is another aspect in need of attention.
In the selected papers, the method achieved information exchange between vehicle and
vehicle/infrastructure by default. However, data dropout is unavoidable in the actual process of
network communication and data transmission, which may greatly affect performance of traffic
signal control system. Unfortunately, our review focuses more on the signal control methods.

7. Conclusions
In this paper, we present a thorough and systematic review on adaptive traffic signal control
in a connected vehicle environment. In order to have a strict evaluation process, this review has
provided a detailed discussion and analysis of adaptive traffic signal control methods, such as the
method implemented in the selected papers, the estimation of unequipped vehicle status, and the
simulation platform employed in those papers. The review has also carefully discussed advantages
and disadvantages of the different methods or strategies used in the selected papers.
To our knowledge, this is the first systematic review of the existing methods of adaptive traffic
signal control in a connected vehicle environment. The existing adaptive signal control methods mainly
focus on two research directions: one is to optimize the signal timing and the other is to optimize
the queue. The best available evidence indicates that adaptive traffic signal control can significantly
reduce the delay and improve the road traffic efficiency. The present systematic review shows that
adaptive traffic signal control research in a connected vehicle environment is in its infancy. Limited by
the development of connected vehicle technology and hardware support, the proposed methods can
only be verified by simulation experiments. Future work examining their adaptability and validity
based on the field testing is warranted. Finally, further research is needed to develop efficient and
generic adaptive traffic signal control methods in a connected vehicle environment.

8. Future Work
Based on the literature review, a thorough analysis of adaptive traffic signal control in a connected
vehicle environment suggests that there are significant opportunities for innovation in adaptive traffic
signal control research within this domain. These include:
Information 2017, 8, 101 21 of 24

• The existing signal control models and optimization methods are based primarily on unsaturated
traffic flow. With the rapid increase in motorization level, the road traffic congestion has become
a common problem all over the world. Although the connected vehicle technology will reduce
the traffic congestion in a certain degree, traffic congestion remains a problem in the period
ahead [60]. Therefore, traffic signal control models and strategies for saturated and over-saturated
intersections are one of the important directions for future research.
• Intelligent control and artificial intelligence technology (such as genetic algorithm, reinforcement
learning, expert system, etc.) provide more choices for the optimization algorithm and signal
control method. Compared with the existing optimization methods, the advantage of intelligent
control is that its control algorithm has a strong approximation nonlinear function without relying
on the precise mathematical model. This may be an effective method for a traffic signal control
system that is hard to build a better mathematical model, especially in a connected vehicle
environment. Although [45,61] proposed the reinforcement learning based signal control methods
and [32] developed a traffic signal control based on approximate dynamic programming, research
on intelligent control in a connected vehicle environment is still limited. Intelligent control will
attract more researchers’ attention.
• Although connected vehicles will have rapid development in the near future, the transit, bus
rapid transit, new tram, and other public transport systems will still have a critical role in the
whole transport system. Public transportation development has attracted increasing attention,
which is the rational trend of cities’ passenger traffic structure. Some authors [46,47] proposed
multi-modal traffic signal control methods, but the research on multi-modal traffic is still limited
at present. Therefore, the special vehicle priority needs to be taken into account in regard to
developing the intersection control methods in a connected vehicle environment.
• In future traffic control system research, the advantages of distributed system, centralized system,
and multilayer distributed system should be fully taken to account. Based on the above methods,
the complex signal controls can be simplified into several logical steps to ultimately achieve global
optimization. The signal control system should be more flexible, switchable, and adaptive to
different control system structure, which can be applicable to different traffic scenarios.
• Autonomous vehicle technology recently has attracted more and more researchers’ attention.
Autonomous vehicles (AVs) represent an emerging transportation mode for driverless
transport [62–65]. Compared to the existing conventional vehicles and present connected
vehicles-based control method, the traffic signal control method based on an autonomous
vehicle environment will undergo a new stage of development. Some innovative ideas have
been proposed based on autonomous vehicles. Some [66,67] proposed the reservation-based
intersection control methods, which allocate “the right of way” of the intersection based on the
series of pre-defined rules. Others [68,69] proposed the trajectory-based algorithms. Under this
scenario, traffic signal controllers at intersections are removed and the intersection control is
based on all vehicle trajectories through the intersection. In summary, there are two main types of
algorithms used to optimize the intersection control in an autonomous vehicle environment: signal
scheduling and trajectory planning. Reservation-based algorithms focus mainly on obtaining
the optimal sequence of each lane by sorting the requests from upcoming AVs. Trajectory-based
algorithms, however, benefit the connectivity of AVs to make preparations for the optimal
departure timing and speed far ahead from the stop line.

Several recommendations put forward in this review provide a foundation for future research on
traffic signal control method in a connected vehicle environment. The relatively recent breakthroughs
in autonomous vehicle technology have allowed researchers to investigate the impact of autonomous
vehicles on intersection signal control. However, it is clear that the application of connected vehicle
technology in the traffic signal control domain is still in its early stages. There remains considerable
opportunities in developing the intersection signal control in a connected vehicle environment.
Information 2017, 8, 101 22 of 24

Conflicts of Interest: The authors declare no conflict of interest.

References
1. Olia, A.; Abdelgawad, H.; Abdulhai, B.; Razavi, S.N. Assessing the Potential Impacts of Connected Vehicles:
Mobility, Environmental, and Safety Perspectives. J. Intell. Transp. Syst. Technol. Plan. Oper. 2014, 23, ix–xii.
[CrossRef]
2. Li, M.; Boriboonsomsin, K.; Wu, G.; Zhang, W.B.; Barth, M. Traffic Energy and Emission Reductions at
Signalized Intersections: A Study of the Benefits of Advanced Driver Information. Int. J. Intell. Transp.
Syst. Res. 2009, 7, 2327–2332.
3. Coelho, M.C.; Farias, T.L.; Rouphail, N.M. Impact of speed control traffic signals on pollutant emissions.
Transp. Res. Part D Transp. Environ. 2005, 10, 323–340. [CrossRef]
4. Denney, R.W., Jr.; Curtis, E.; Olson, P. The National Traffic Signal Report Card. ITE J. 2012, 86, 22–26.
5. Fatima, K.; Fatima, K. Modal Congestion Management Strategies and the Influence on Operating
Characteristics of Urban Corridor. Master Thesis, RMIT University, Melbourne, Australia, 2015.
6. Auditorgeneral, V.; Office, S. Managing Traffic Congestion; Highway Traffic Control: Saugatuck, CT, USA, 2013.
7. Maslekar, N.; Mouzna, J.; Boussedjra, M.; Labiod, H. CATS: An adaptive traffic signal system based on
car-to-car communication. J. Netw. Comput. Appl. 2013, 36, 1308–1315. [CrossRef]
8. Hoogendoorn, S.; Knoop, V. Traffic Flow Theory and Modelling; Edward Elgar Publishing Limited: Cheltenham,
UK, 2012.
9. Zheng, X.; Recker, W.; Chu, L. Optimization of Control Parameters for Adaptive Traffic-Actuated Signal
Control. J. Intell. Transp. Syst. 2010, 14, 95–108. [CrossRef]
10. Zheng, X.; Chu, L. Optimal Parameter Settings for Adaptive Traffic-Actuated Signal Control. In Proceedings
of the International IEEE Conference on Intelligent Transportation Systems, Beijing, China, 12–15
October 2008.
11. Sims, A.G.; Dobinson, K.W. The Sydney coordinated adaptive traffic (SCAT) system philosophy and benefits.
IEEE Trans. Veh. Technol. 1980, 29, 130–137. [CrossRef]
12. Gartner, N.H. OPAC: A Demand Responsive Strategy for Traffic Signal Control; Transportation Research Record:
Washington, DC, USA, 1983; No. 906; pp. 75–81.
13. Bing, B.; Carter, A. SCOOT: The World’s Foremost Adaptive Traffic Control System; Traffic Technology
International’95; UK and International Press: Surrey, UK, 1995.
14. Mirchandani, P.; Head, L. A real-time traffic signal control system: Architecture, algorithms, and analysis.
Transp. Res. Part C Emerg. Technol. 2001, 9, 415–432. [CrossRef]
15. Henry, J.J.; Farges, J.L.; Tuffal, J. The Prodyn Real Time Traffic Algorithm. Control Transp. Syst. 1984, 16,
305–310.
16. Brilon, W.; Wietholt, T. Experiences with Adaptive Signal Control in Germany. Transp. Res. Rec. J. Transp. Res.
Board 2013, 2356, 9–16. [CrossRef]
17. Feng, Y.; Head, K.L.; Khoshmagham, S.; Zamanipour, M. A real-time adaptive signal control in a connected
vehicle environment. Transp. Res. Part C Emerg. Technol. 2015, 55, 460–473. [CrossRef]
18. Huang, Q.; Miller, R. Reliable Wireless Traffic Signal Protocols for Smart Intersections. In Proceedings of the
14th ITS America Annual Meeting and Exposition, San Antonio, TX, USA, 26–28 April 2004.
19. Wu, J.; Abbas-Turki, A.; Correia, A.; El Moudni, A. Discrete Intersection Signal Control. In Proceedings of
the IEEE International Conference on Service Operations and Logistics, and Informatics, Philadelphia, PA,
USA, 27–29 August 2007.
20. Cheng, J.; Wu, W.; Cao, J.; Li, K. Fuzzy Group Based Intersection Control via Vehicular Networks for Smart
Transportations. IEEE Trans. Ind. Inform. 2017, 13, 751–758. [CrossRef]
21. Ahmane, M.; Abbas-Turki, A.; Perronnet, F.; Wu, J.; El Moudni, A.; Buisson, J.; Zeo, R. Modeling and
controlling an isolated urban intersection based on cooperative vehicles. Tramsp. Res. Part C Emerg. Technol.
2013, 28, 44–62. [CrossRef]
22. Tiaprasert, K.; Zhang, Y.; Wang, X.B.; Zeng, X. Queue Length Estimation Using Connected Vehicle Technology
for Adaptive Signal Control. IEEE Trans. Intell. Transp. Syst. 2015, 16, 2129–2140. [CrossRef]
23. Guler, S.I.; Menendez, M.; Meier, L. Using connected vehicle technology to improve the efficiency of
intersections. Transp. Res. Part C Emerg. Technol. 2014, 46, 121–131. [CrossRef]
Information 2017, 8, 101 23 of 24

24. Lee, J.; Park, B.; Yun, I. Cumulative Travel-Time Responsive Real-Time Intersection Control Algorithm in the
Connected Vehicle Environment. J. Transp. Eng. 2013, 139, 1020–1029. [CrossRef]
25. Vrabel, M. Preferred Reporting Items for Systematic Reviews and Meta-Analyses. Oncol. Nurs. Forum 2015,
42, 552. [CrossRef] [PubMed]
26. Peirce, S.; Mauri, R. Vehicle-Infrastructure Integration (VII) Initiative Benefit-Cost Analysis: Pre-Testing Estimates;
Draft Report; Intelligent Transportation Systems Joint Program Office: Washington, DC, USA, 2007.
27. Pandit, K.; Ghosal, D.; Zhang, H.M.; Chuah, C.N. Adaptive Traffic Signal Control with Vehicular Ad hoc
Networks. IEEE Trans. Veh. Technol. 2013, 62, 1459–1471. [CrossRef]
28. Gradinescu, V.; Gorgorin, C.; Diaconescu, R.; Cristea, V.; Iftode, L. Adaptive Traffic Lights Using Car-to-Car
Communication. In Proceedings of the IEEE Vehicular Technology Conference—VTC 2007-Spring, Dublin,
Ireland, 22–25 April 2007.
29. Webster, F.V.; Cobbe, B.M. Traffic Signals; Road Research Technical Paper No. 56; Her Majesty’s Stationery
Office: London, UK, 1966; Volume 4, pp. 206–207.
30. Howard, R.A. Dynamic Programming. Manag. Sci. 1966, 12, 317–348. [CrossRef]
31. Werbos, P.J. Approximate dynamic programming for real-time control and neural modeling. In Handbook of
Intelligent Control: Neural Fuzzy & Adaptive Approaches; Van Nostrand Reinhold: New York, NY, USA, 1992.
32. Cai, C.; Wang, Y.; Geers, G. Vehicle-to-infrastructure communication-based adaptive traffic signal control.
IET Intell. Transp. Syst. 2013, 7, 351–360. [CrossRef]
33. Chang, H.J.; Park, G.T. A study on traffic signal control at signalized intersections in vehicular ad hoc
networks. Ad Hoc Netw. 2013, 11, 2115–2124. [CrossRef]
34. Younes, M.B.; Boukerche, A. Intelligent Traffic Light Controlling Algorithms Using Vehicular Networks.
IEEE Trans. Veh. Technol. 2016, 65, 5887–5899. [CrossRef]
35. Nafi, N.S.; Khan, J.Y. A VANET based Intelligent Road Traffic Signalling System. In Proceedings of the
Telecommunication Networks and Applications Conference, Brisbane, QLD, Australia, 7–9 November 2012.
36. Jennings, N.R.; Sycara, K.; Wooldridge, M. A Roadmap of Agent Research and Development. Auton. Agents
Multi-Agent Syst. 1998, 1, 7–38. [CrossRef]
37. Kari, D.; Wu, G.; Barth, M.J. Development of an agent-based online adaptive signal control strategy
using connected vehicle technology. In Proceedings of the IEEE International Conference on Intelligent
Transportation Systems, Qingdao, China, 8–11 October 2014.
38. Chou, L.D.; Deng, B.T.; Li, D.C.; Kuo, K.W. A passenger-based adaptive traffic signal control
mechanism in Intelligent Transportation Systems. In Proceedings of the International Conference on ITS
Telecommunications, Taipei, Taiwan, 5–8 November 2012.
39. Priemer, C.; Friedrich, B. A decentralized adaptive traffic signal control using V2I communication data.
In Proceedings of the International IEEE Conference on Intelligent Transportation Systems, St. Louis, MO,
USA, 4–7 October 2009.
40. Priemer, C.; Friedrich, B. A method for tailback approximation via C2I-data based on partial penetration.
In Proceedings of the 15th World Congress on Intelligent Transport Systems and ITS America’s 2008 Annual
Meeting, New York, NY, USA, 16–20 November 2008.
41. Wallace, C.E.; Courage, K.G.; Reaves, D.P.; Schoene, G.W.; Euler, G.W. TRANSYT-7F User’s Manual; Federal
Highway Administration: Washington, DC, USA, 1984.
42. Goodall, N.J.; Smith, B.L.; Park, B. Traffic Signal Control with Connected Vehicles. Transp. Res. Rec. J. Transp.
Res. Board 2013, 2381, 65–72. [CrossRef]
43. Shaghaghi, E.; Jabbarpour, M.R.; Noor, R.M.; Yeo, H.; Jung, J.J. Adaptive green traffic signal controlling using
vehicular communication. Front. Inf. Technol. Electr. Eng. 2017, 18, 373–393. [CrossRef]
44. Islam, S.M.A.B.; Hajbabaie, A. Distributed coordination and optimization for signal timing in connected
transportation networks. Transp. Res. Part C Emerg. Technol. 2017, 80, 272–285. [CrossRef]
45. Liu, W.; Qin, G.; He, Y.; Jiang, F. Distributed Cooperative Reinforcement Learning-Based Traffic Signal
Control that Integrates V2X Networks’ Dynamic Clustering. IEEE Trans. Veh. Technol. 2017. [CrossRef]
46. He, Q.; Head, K.L.; Ding, J. PAMSCOD: Platoon-based Arterial Multi-modal Signal Control with Online
Data. Transp. Res. Part C 2012, 20, 164–184. [CrossRef]
47. He, Q.; Head, K.L.; Ding, J. Multi-modal traffic signal control with priority, signal actuation and coordination.
Transp. Res. Part C Emerg. Technol. 2014, 46, 65–82. [CrossRef]
Information 2017, 8, 101 24 of 24

48. Hu, J.; Park, B.B.; Lee, Y.J. Coordinated transit signal priority supporting transit progression under Connected
Vehicle Technology. Transp. Res. Part C 2015, 55, 393–408. [CrossRef]
49. Ezawa, H.; Mukai, N. Adaptive Traffic Signal Control Based on Vehicle Route Sharing by Wireless Communication;
Springer: Berlin/Heidelberg, Germany, 2010; pp. 280–289.
50. Xiang, J.; Chen, Z. An adaptive traffic signal coordination optimization method based on vehicle-to-
infrastructure communication. Cluster Comput. 2016, 19, 1–12. [CrossRef]
51. Tomescu, O.; Moise, I.M.; Stanciu, A.E.; Batros, I. Adaptive Traffic Light Control System Using Ad Hoc
Vehicular Communications Network. UPB Sci. Bull. 2012, 74, 67–78.
52. Hobeika, A.G.; Kim, T. Assessment of certain applications of Vehicle-to-Vehicle communication in an urban
network. Wit Trans. Built Environ. 2014, 138, 405–417.
53. Sommer, C.; German, R.; Dressler, F. Bidirectionally Coupled Network and Road Traffic Simulation for
Improved IVC Analysis. IEEE Trans. Mob. Comput. 2010, 10, 3–15. [CrossRef]
54. PTV Group. VISSIM 5.40: User Manual; PTV Group: Karlsruhe, Germany, 2011.
55. Krajzewicz, D.; Erdmann, J.; Behrisch, M.; Bieker, L. Recent Development and Applications of
SUMO—Simulation of Urban MObility. Int. J. Adv. Syst. Meas. 2012, 5, 128–138.
56. Wang, S.Y.; Lin, C.C. NCTUns 5.0: A Network Simulator for IEEE 802.11(p) and 1609 Wireless Vehicular
Network Researches. In Proceedings of the Vehicular Technology Conference—VTC 2008-Fall, Calgary, AB,
Canada, 21–24 September 2008.
57. Manual AIMSUN User. 6.1: Microsimulator and Mesosimulator in AIMSUN; Transport Simulation Systems:
Barcelona, Spain, 2009.
58. Wiering, M.; Vreeken, J.; Van Veenen, J.; Koopman, A. Simulation and optimization of traffic in a city.
In Proceedings of the Intelligent Vehicles Symposium, Parma, Italy, 14–17 June 2004.
59. Guide, MATLAB User. The Mathworks; MathWorks Inc.: Natick, MA, USA, 1998.
60. Jabbarpour, M.R.; Zarrabi, H.; Khokhar, R.H.; Shamshirband, S.; Choo, K.K.R. Applications of computational
intelligence in vehicle traffic congestion problem: A survey. Soft Comput. 2017. [CrossRef]
61. Yang, K. A Reinforcement Learning Based Traffic Signal Control Algorithm in a Connected Vehicle
Environment. In Proceedings of the 17th Swiss Transport Research Conference (STRC 2017), Ascona,
Switzerland, 17–19 May 2017.
62. Fagnant, D.J.; Kockelman, K.M. Dynamic Ride-Sharing and Optimal Fleet Sizing for a System of Shared
Autonomous Vehicles. In Proceedings of the Transportation Research Board Annual Meeting, Washington,
DC, USA, 11–15 January 2015.
63. Fagnant, D.J.; Kockelman, K.M. Dynamic ride-sharing and fleet sizing for a system of shared autonomous
vehicles in Austin, Texas. Transportation 2016. [CrossRef]
64. Truong, L.T.; De Gruyter, C.; Currie, G.; Delbosc, A. Estimating the trip generation impacts of autonomous
vehicles on car travel in Victoria, Australia. Transportation 2017, 1–14. [CrossRef]
65. Woodard, M.; Sedigh, S. Modeling of Autonomous Vehicle Operation in Intelligent Transportation Systems.
In Proceedings of the International Workshop on Software Engineering for Resilient Systems, Kiev, Ukraine,
3–4 October 2013.
66. Dresner, K.; Stone, P. Multiagent Traffic Management: A Reservation-Based Intersection Control Mechanism.
In Proceedings of the International Joint Conference on Autonomous Agents and Multiagent Systems,
Utrecht, The Netherlands, 25–29 July 2005.
67. Hausknecht, M.; Au, T.C.; Stone, P. Autonomous Intersection Management: Multi-intersection optimization.
In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, San Francisco,
CA, USA, 25–30 September 2011.
68. Lee, J.; Park, B. Development and Evaluation of a Cooperative Vehicle Intersection Control Algorithm under
the Connected Vehicles Environment. IEEE Trans. Intell. Transp. Syst. 2012, 13, 81–90. [CrossRef]
69. Li, Z.; Elefteriadou, L.; Ranka, S. Signal control optimization for automated vehicles at isolated signalized
intersections. Trans. Res. Part C 2014, 49, 1–18. [CrossRef]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).