Intelligent Traffic Management: A Review of Challenges, Solutions, and Future Perspectives

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/351126263
Intelligent Trafﬁc Management: A Review of Challenges, Solutions, and

Future Perspectives
Article in Transport and Telecommunication · April 2021

DOI: 10.2478/ttj-2021-0013
CITATIONS READS
6 4,314
2 authors, including:
Roopa Ravish
PESUniversity
13 PUBLICATIONS 27 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Intelligent Transportation System View project
Software Implementation of an Automatic Movable Road Barrier View project
All content following this page was uploaded by Roopa Ravish on 22 May 2021.
The user has requested enhancement of the downloaded file.

Transport and Telecommunication Vol. 22, no.2, 2021
Transport and Telecommunication, 2021, volume 22, no. 2, 163–182

Transport and Telecommunication Institute, Lomonosova 1, Riga, LV-1019, Latvia
DOI 10.2478/ttj-2021-0013
INTELLIGENT TRAFFIC MANAGEMENT: A REVIEW OF

CHALLENGES, SOLUTIONS, AND FUTURE PERSPECTIVES
Roopa Ravish1, Shanta Ranga Swamy2
1
Department of Computer Science and Engineering, PES University,
Bangalore, India
rooparavish@pes.edu
2
Department of Computer Science and Engineering, Rashtreeya Vidyalaya College of Engineering,
Bangalore, India
shantharangaswamy@rvce.edu.in
Recent years have witnessed a colossal increase of vehicles on the roads; unfortunately, the infrastructure of roads and
traffic systems has not kept pace with this growth, resulting in inefficient traffic management. Owing to this imbalance, traffic jams
on roads, congestions, and pollution have shown a marked increase. The management of growing traffic is a major issue across the
world. Intelligent Transportation Systems (ITS) have a great potential in offering solutions to such issues by using novel
technologies. In this review, the ITS-based solutions for traffic management and control have been categorized as traffic data
collection solutions, traffic management solutions, congestion avoidance solutions, and travel time prediction solutions. The
solutions have been presented along with their underlying technologies, advantages, and drawbacks. First, important solutions for
collecting traffic-related data and road conditions are discussed. Next, ITS solutions for the effective management of traffic are
presented. Third, key strategies based on machine learning and computational intelligence for avoiding congestion are outlined.
Fourth, important solutions for accurately predicting travel time are presented. Finally, avenues for future work in these areas
are discussed.
Keywords: Congestion Management, Computational Intelligence, Intelligent Transport Systems, Intelligent Traffic Control, Traffic
Data Collection, Traffic Management, Travel Time Prediction
1. Introduction
Congestion of traffic is a key problem faced in a majority of metro cities, especially in the
developing world. Traffic congestion comprises of queues, reduced speeds, and increased travel
durations, which fatigues commuters and results in stress, thereby bringing down productivity and
creating intangible societal expenses (Emo et al., 2016). In addition, different factors, such as use of
natural resources, surroundings, commuter safety, etc., are affected in either a direct or an indirect
manner. Thus, congestion of traffic poses a challenge to every growing city. With the increasing
congestion, the construction of new roads is seldom fruitful because of various circumstances. Contrarily,
constructing new roads could often add to congestion by creating more demands for motor travel, thereby
depleting the added capacity rapidly (Pfleiderer and Dieterich, 1995). This is getting increasingly evident
as congestions with delays are being noticed in big cities. Thus, several mitigation measures are being
implemented through the years.
The increasing challenges associated with traffic congestion attracted the attention of
academicians, who have proposed numerous techniques for managing traffic and solving the congestion
issue. Such solutions range from the latest data collecting equipment with wireless sensor network (WSN)
protocols (Faye et al., 2012; Collotta et al., 2014) to route schemes (Chen et al., 2012; Bell et al., 2012)
and traffic forecasting methods (Min and Wynter, 2011; Shen and Wynter, 2012). Several researchers
were interested in the application of Intelligent Transportation Systems (ITS) to manage traffic
management issues. Liang and Wakahara (2013) proposed city traffic prediction models using varied
modeling methods, one based on the network traffic flow propagation, and the other based on the time-
varying spare flow volume on the related road link. This paper investigates the technologies directed
towards developing novel systems that address many of the traffic congestion problems. The findings of
these researchers indicated that roads could be safer and greener with enhanced efficiency through the
application of emerging ITS technologies (Sładkowski and Pamuła, 2016). However, several challenges
are encountered during the implementation of ITS.
163
Studies pointed out that effective control and management of traffic and accurate prediction of
travel time are the two major challenges faced in the application of ITS. However, despite the emphasis
on these challenges, only a few researchers have rendered solutions based on the functionality of ITS
(Van Arem et al., 1997; Papageorgio et al., 2003; Barros et al., 2015; Chen and Chen, 2019). In addition,
only few papers have compared the benefits and shortcomings of the suggested solutions (Chhatpar et al.,
2018; De Souza et al., 2017; Avatefipour and Sadry, 2018). Therefore, in the current research paper, our
objectives are as follows:
1) To present some important state-of-the-art ITS solutions for managing traffic congestion
2) To categorize ITS, based on their functionalities with working principles and underlying
technologies
3) To provide insights on the drawbacks and benefits of the current ITS proposals and help
stakeholders to select the technology for adoption
4) To provide future directions
Having conducted an in-depth survey of previous empirical studies in this area, the outcomes of
this study contributes to the current knowledge of traffic management by providing a novel perspective on
its functional constructs. This research also contributes to the existing literature by rendering critical
insights into the pros and cons of the existing systems, thereby aiding decision makers in the selection of
technologies for their cities.
The rest of the paper is organized as follows: the first section deals with the process of this review,
the next section deals with effective traffic control and management by using ITS, and the subsequent
subsections deal with traffic data collection, congestion management, congestion avoidance, and accurate
prediction of travel time. Finally, the conclusions and directions for future work are discussed.
2. Review Process
A detailed classification and analysis of literature was considered as a part of the current research.
The following methodology was adopted:
1) The database was updated by selecting the latest literature. The collected literature was
reviewed until June 2019.
2) Both hard copies in reputed local libraries and soft copies from the Internet were accessed for
the literature review.
3) Popular search engines, such as www.google.com, www.altavista.com, etc., were employed to
gather the subject-related literature from a multitude of sources. Although it has been attempted
to include as many relevant works as possible, this list is by no means either complete or
exhaustive.
4) The classification scheme was developed by looking at the nature of the studies. Later, the
studies were examined to detect commonalities, content, advantages, and disadvantages.
5) Finally, the studies were examined with a view to suggest future avenues for research.
3. Effective Traffic Control and Management using ITS
ITS attempts to blend network-related data, such as vehicular cloud computing (Shojafar et al.,
2016), WSN (Cordeschi et al., 2015), and other approaches, in a seamless manner for managing traffic
effectively. The traffic congestion handling solution by ITS can be functionally categorized as traffic data
collection, traffic management, congestion avoidance, and travel time prediction (Figure 1). The
subsequent subsections provide the review of studies that examined the techniques used for delivering
these functions.
Figure 1. Functional categories for congestion management solutions by using ITS
164
3.1. Traffic Data with Travel Time Constraints

Collection of traffic data is one of the primary and important functions in managing the traffic as
all other traffic management procedures rely on the accurate prediction of traffic. Several researchers
proposed numerous methods to collect the data related to traffic and road conditions. One of the common
methods used for the collection of data is the use of sensors fixed at a specific location. Sen et al. (2009)
developed a data collection system by placing a microphone on the roadside to record the sounds of
vehicular honks for estimating the congestion level. However, error in the accuracy of sensing congestion
during multiple and simultaneous vehicular honking restricted this technique. Na et al. (2015) modified
the technology by using a module comprising of 37 microphones. Despite the fair outcomes in bad
weathers and poor light, its high cost and limitations in determining congestion accuracy during high
vehicular movement limited its use. The development of machine learning and artificial intelligence
changed the design of sensors and Fan et al. (2018) developed a sensor for collecting the data to detect
vehicular congestion by using machine-learning technology. These sensors were placed at various toll
collection centers. The findings revealed better results when compared to other sensors. However, higher
costs limited its application.
Adaptive architecture framework was another technology proposed by Khan (2017) to gather
traffic data. This technology focused on detecting bus traffic. The author stated that the selection of
private or public mode of transportation depends upon one’s lifestyle and it is the responsibility of ITS to
balance both. However, the use of private vehicles has a direct impact on congestion. If potential public
transport issues are solved, then a perfect balance could be achieved to build software, such as Global
Positioning System (GPS), Connected Vehicle (CV), Sensor-based, and Internet of Vehicles (IoV).
Increase in the usage of smart devices and vast improvements in wireless technology have facilitated the
application of GPS-based wireless techniques and the usage of positional coordinates for collecting
vehicle-related data.
Haferkamp et al. (2017) proposed a vehicular data collection technique using wireless networks.
Received Signal Strength Indicator (RSSI) was used to detect vehicles and classify vehicle types. RSSI
performed well with high detection accuracy. However, the antenna-related expenses restricted its
application to a certain extent. Bernas (2018) updated the wireless network-based vehicle detection
systems by using more transmitters and receivers to gain more data with higher accuracy. Despite its good
performance, the application of this system was restricted due to high expenses. Still, these wireless-based
data collection and vehicle detection systems are largely considered as optimal to control traffic
congestion. In addition to these techniques, Ghosh et al. (2015) proposed seismometer-based and
geophone-based technology to detect vehicles. However, the fall of detection window prior to the event
completion was considered as its prime drawback, which in turn restricted its application. Furthermore,
Rivas et al. (2017) proposed a piezoelectric technologies-based technique that had drawbacks, such as
high susceptibility to weather conditions and low efficiency.
In brief, it can be pointed out that despite the utilization of numerous techniques to collect data and
detect vehicles, the application of such techniques relied on the specific nature of the location where the
traffic has to be managed. Hence, the design of the data collection module largely depends on the
requirement of the specific locations. Table 1 summarizes the technologies used in collecting traffic-
related data.
Table 1. Summary of technologies adopted for gathering traffic data
Reference Technology Principles Advantages Disadvantages
Use of Data Mining
No changes in the Influx of new data sources
techniques and
Adaptive architecture existing system needed. will impact whole life cycle
Khan (2017) Machine Learning
framework Real-time monitoring of data and performance of
algorithms for traffic
of buses and routes data acquisition
monitoring
Vehicles are equipped Accuracy and reliability
Fan et al. Machine learning and
with sensors which Simple and economical issues when traffic volumes
(2018) big data
provide real-time data and density fluctuate
Economical, can be Error in results as vehicle
Sen et al. Sound recording Assessment of Doppler
used where traffic is speeds increase, overlapping
(2009) (microphones) shift of vehicular honks
unorganized honks will affect results
The developed Nericell
virtually reorients the Privacy concerns of
Energy efficient,
Mohan et al. GPS, accelerometer, accelerometer on a smartphone users, getting
suitable for bad roads,
(2008) and microphone smartphone, and users to participate in the
chaotic traffic
detects honks and does shared system
localization.
165
Reference Technology Principles Advantages Disadvantages

RSSI information is
Wireless network with
Haferkamp used for detecting High classifying The directional antennae
three transmitters and
et al. (2017) vehicles and classifying accuracy utilized makes it expensive.
receivers
vehicle type.
RSSI information is High classifying
Wireless network using The utilized directional
Bernas et al. used for detecting accuracy, low energy
four transmitters and antennae makes it
(2018) vehicles and classifying consumption, easy
receivers expensive.
vehicle type. installation
Algorithms analyze the
Rivas et al. Piezoelectric amplitudes and Economical, high Weather conditions could
(2017) acceleration sensors frequency range of accuracy affect efficiency
vibrations
Pseudo-Wigner–Ville
CFAR threshold increases
Distribution assisted
as CPA is approached,
Ghosh et al. Seismometer and Rényi entropy method Minimal false alarms,
causing fall of detection
(2015) geophone with an enhanced good detection
window prior to event
CFAR detector for
completion
seismic waves
Thirty seven- Can function even in
Na et al. High costs of equipment,
microphone poor light and adverse
(2015) accuracy limitations
arrangement weather
3.2. Traffic Management Systems

Traffic Management System (TMS) is concerned with the organizing, arranging, guiding, and
controlling of traffic, both moving and stationary, along with other vehicles, cyclists, and pedestrians.
TMS aims to ensure safety, orderliness, and efficiency during the movement of people and goods, and
enhance the quality of environment around traffic locations (Underwood, 1990). The present section deals
with ITS solutions for effectively managing traffic and the drawbacks associated with their functioning.
WSNs are being increasingly used for detecting traffic and avoiding congestion. WSNs are
effective because of their fast data transfer, simple installation, minimal maintenance, small size, and
cost-effectiveness, when compared with other network choices. Considerable research has been
conducted on TMS by using WSNs for avoiding congestion, enabling priority for emergency vans, and
reducing waiting time of vehicles at signals.
Friesen et al. (2004) proposed a comprehensive data collection mechanism. This system uses
different WSN devices and technologies for gathering information at the main traffic signals in urban
areas. Vehicular count and trajectory data are gathered by using a Bluetooth mechanism, and this data is
forwarded to a master node that handles all received data through an IEEE 802.15.4 protocol. The real-
time data is sent by the node to a server at a five-minute interval through a cellular phone having GSM.
This data is saved in the database and retrieved by users to obtain the required real-time data. It is
evidenced that Bluetooth tracking is a reliable way of monitoring traffic and vehicles in real time.
However, this method suffered low data rate and short-range problems along with severe security issues.
Saqib and Lee (2010) proposed an application, which estimated the vehicular location and speed
by employing WSNs. Anchor sensing nodes were employed as pavement readers. As soon as a moving
vehicle came within the operational range of the nodes, the positional signal was transmitted by using a
symmetrical two-sided range algorithm. The speed of the vehicles is estimated by measuring the locations
at varied intervals. This suggested model was found to be more economical than other models for
detecting the location and speed of vehicles. However, increased error rate during higher traffic flow
restricted its usage.
A user-dependent traffic data-gathering strategy relying on WSNs was suggested by Zhou et al.
(2013) in a study for achieving a compromise between the sensor power consumed and the delay in
delivering data by employing every routing feature for decision-making. However, security issues and
low data rate were its major drawbacks, which limited its scope.
Ahmad et al. (2013) suggested a traffic management technique by using a channel switching
approach for reducing communication expenses and power consumption. Simulation for this suggested
model was conducted by using OPNET. It was seen that that the model could be feasibly implemented at
traffic signals. However, the proposed solution failed due to the interference of vehicle-to-vehicle
channels.
Bruno and Nurchis (2013) developed two useful data gathering algorithms: Probabilistic Data
Collection (PDC) for vehicle multimedia SNs and GREEDY. The researchers simulated this algorithm by
using VANET MobiSim and NS-2 simulator. The output indicated that the GREEDY algorithm could
166
cover more uniformly while consuming reduced network bandwidth. However, these techniques failed to
consider cooperative storage among vehicles and image replications. This in turn could significantly
affect the accuracy of data collection and hence this proposal was not much appreciated.
Another technology used in managing the traffic is RFID-based intelligent system. Chao and Chen
(2014) presented a RFID-based intelligent TMS for determining traffic movement. The proposed
intelligent traffic light control system (ITLCS) uses an RFID system, which complies with the IEEE
802.11p protocol for detecting the number of vehicles and finding the time spent by vehicles on main
roads and on side roads, passing through the intersection throughout the period of a green light. They used
Zig Bee modules to send real time data like weather conditions and the vehicle registration information to
the regional control center. The proposed system could perform remote transmission and reduce traffic
accidents. However, despite a fair attempt, the scope of using Zig Bee for data transfer was restricted
because Zig Bee always displayed a very low data transfer rate that could hamper accuracy.
Another technique named as optimal VSN routing strategy was proposed by Choi et al. (2015) in
an attempt to design a traffic management system. The researchers studied the factors that influence the
delay performance in VSNs. They formulated the packet routing problem as a Markov Decision Process
(MDP) and developed an Optimal VSN Data Forwarding (OVDF) algorithm. They conducted simulations
on a GloMoSim simulator. They compared the performance of OVDF with the Vehicle Assisted Data
Delivery (VADD) algorithm and Trajectory Based Data (TBD) forwarding scheme. The simulation
results demonstrated that the OVDF gave the best sensing coverage and a higher packet delivery ratio,
and reduced the delay about 25% and 20% when compared to VADD and TBD, respectively. Despite
such high performances, increased error during higher traffic casted some shadow on its applicability.
In a recent study, Barthélemy et al. (2019) utilized edge-computing technologies. The study
yielded satisfactory results in managing the traffic with advantages, such as multi-modal detection,
privacy compliance, scalability, and interoperability. However, certain operational issues in bad weather
and low light constituted its drawbacks. The strategies used for traffic management are summarized in
Table 2.
Table 2. Summary of technologies used for managing traffic
Authors Technology Principle Advantages Disadvantages

Bruno and Multiple mobile Use of Probabilistic Data Coverage more uniform, Does not consider
Nurchis (2013) cameras Collection and GREEDY consumes less network cooperative storage
algorithm for collecting bandwidth among vehicles and image
data, and analysis of replications
collected images
Chao and Chen IRFID system, RFID detects traffic flow, Performs remote Zigbee systems have low
(2014) ZigBee wireless and a sensor analyzes data transmission and reduces data transfer rates that
network for traffic control using an traffic accidents, power could hamper accuracy
communication extension algorithm savings, and scalable.
Saqib and Lee WSN Two anchor nodes placed Economical, robust, and Errors when traffic is
(2010) at fixed distance. A requires lesser heavy
symmetric double-sided computation
two-way ranging
algorithm gives location
of vehicle
Choi et al. (2015) Markov Decision A road network graph Highr packet delivery Errors when traffic is
Process and model captures the vehicle ratio, reduced delay heavy
Optimal VSN effects having
Data Forwarding predetermined trajectories
and creates an easy
routing problem
formulation
Friesen et al. WSN Slave probes transmit Monitoring traffic in real Low data rate, short range,
(2004) traffic data to a master time using Bluetooth security issues
node using XBee
communication, and is
forwarded to a central
server by a GSM network
Zhou et al. WSN Three sensor nodes collect Balances sensor power Low data rate, short range,
(2013) and deliver data. A user- consumed and data delay security issues
driven data-centric duration
strategy for routing
considers multi-routing
data for taking decisions
for meeting various user
needs
167
Authors Technology Principle Advantages Disadvantages

Ahmad et al. WSN Nodes are located in Reduces communication Prone to intra-channel
(2013) vehicles which interact expenses and power interference, security
with road side receivers consumed issues
that transmit the data to
coordinating modules. A
robust channel switching
method minimizes the
energy consumed,
response duration, and
connecting delays
Barthélemy et al. Edge An edge-computing Multi-modal detection, Operational issues in low
(2019) computation equipment uses deep privacy compliance, light and bad weather
neural networks and scalable, and interoperable
computer vision for
tracking traffic in real-
time. The interoperable
Agnosticity framework
gathers, stores, and
manages data from
different sensors
Won et al. (2019) Wi-Fi Two WiFi transceivers Reduced energy Not suitable where traffic
capture data. A consumption, minimal is heavy
convolutional neural connectivity delay and
network and principal reduced interference from
component analysis based nearby intersections
algorithm captures the
vehicles in the data
3.3. Congestion Avoidance Schemes

Traffic congestion occurs when road usage increases. Many strategies and methods, which are
based on machine learning and computational intelligence, have been suggested to prevent congestion.
The key studies are presented here.
Managing Vehicular Networks with Machine learning
Often, effective rerouting strategies are necessary when traffic networks are congested. This section
describes some key approaches proposed in the literature, which use machine-learning approaches.
Araújo et al. (2014) proposed a cooperative vehicular traffic congestion identification and minimization
scheme for reducing traffic congestion. The study employed vehicle-to-vehicle communication and the data
was gathered from all vehicles and processed by fuzzy logic. The paths of vehicles were modified after the
probability of congestion was identified, resulting in the minimization of congestion. However, the major
drawback was that the technique failed to detect accidents, which can be a prime cause for congestion.
Meneguette et al. (2015) proposed an urban congestion identifying technique for reducing traffic
jams by using vehicle-to-vehicle communication along with Artificial Neural Networks (ANN). The
algorithm detected and classified the levels of congestion on the streets. The ANN processed the velocity
and flow density as inputs and returned the congestion level as output, which was transmitted to all
vehicles. However, the technique lacked accuracy during high traffic and did not detect accidents.
De Souza et al. (2015) proposed the application of cooperative rerouting algorithm that runs on K-
Nearest Neighbor for preventing congestion and improving traffic by using vehicle-to-infrastructure
communication. The algorithm detected congestion, classified the traffic, and suggested an alternate
route. However, the technique displayed poor traffic balance, which in turn restricted its scope.
De Souza et al. (2016) proposed an algorithm for avoiding congestion by using a traffic classifying
method with rerouting called CHIMERA. It used vehicle-to-infrastructure communication, along with the
probabilistic K-Shortest Path method and worked out other routes for users. Although the technique
reduced the average trip time, stop time, and distance, its scope was hindered by poor traffic balance.
Ma et al. (2017) proposed the Convolution Neural Networking (CNN) algorithm to identify
congestion and manage traffic. Despite a satisfactory performance in large networks with high accuracy,
the application of CNN was restricted due to the lag in training time, which caused delay in detection of
congestion.
As a means to overcome the drawbacks of CNN in detecting the congestion, Chung and Sohn
(2018) proposed deep CNN model. The model used supervised learning (deep convolution neural
network) for counting the number of vehicles on a strip of road, based on video imagery. The method
168
collectively counted the vehicles in a human-like manner. The findings revealed that the proposed deep
CNN provided congestion reports with fair accuracy and smaller time lag. However, deep CNN could not
distinguish between moving vehicles and stationary vehicles.
Perez-Murueta et al. (2019) proposed the application of another deep learning technique for
detecting traffic congestion. The authors proposed deep learning and k-shortest path algorithm to detect
traffic congestion. However, the findings revealed that the proposed technique could not scale the
intensity of traffic.
Table 3. Summary of machine learning approaches to mitigate congestion
Reference Algorithm Principle Advantages Disadvantages

Chowdhury PROPHET Data from node of adjacent road is subjected Accurate results, Heavy traffic and
et al. (2018) to Time Series analysis, where node easy installation bad weather could
information contains the traffic density of all affect results.
incoming lanes of an intersection (node). Sensors are
Prediction done by using PROPHET expensive
algorithm
Araújo et al. Fuzzy logic For effective identification of traffic Effective No accident
(2014) congestion, an algorithm CARTIM is used congestion detection, limited
that uses a fuzzy logic-based system. The identification and awareness of traffic
method effectively employs the collaborative minimization conditions
communicating channel, avoiding
overloading
Meneguette Artificial The solution intelligent protocol of No accident
et al. (2014) Neural congestion detection (INCIDENT) uses detection, not very
Networks ANN for (i) identifying the various levels of effective for heavy
(ANN) congestion; (ii) categorizing these congestion traffic
levels, and (iii) proposing new paths for
avoiding congestion
De Souza et al. k-Nearest Authors proposed the congestion avoidance Reduces average Traffic balance is
(2016) Neighbour through a traffic classification mechanism trip time, stop poor
(KNN) and a re-routing algorithm (CHIMERA). The time, and distance
k-NN algorithm used by CHIMERA utilizes
the density of vehicles and average road
speed as inputs, and detects congestions
De Souza et al. KNN Same as above, except that a cooperative Reduces average Scalability issues,
(2015) algorithm suggests an alternative path trip time, stop sensitive to noisy
time, and distance, data
better traffic
balance.
Ma et al. CNN Spatio-temporal vehicular dynamics are Works well in Training time is
(2017) transformed to imagery detailing the large scale more
time/space relations of flowing traffic by 2-D networks, good
time–space matrices. CNN extracts abstract accuracy
traffic characteristics and predicts traffic
speed in the network
Hong (2011) Support Vector Model integrates the seasonal support vector Accurate forecasts Long response time,
regression model and chaotic simulated considerable
annealing algorithm for predicting traffic computer memory
flow consumed on
training
Ke et al. Connected Parameters for traffic flow, e.g. speed, Accurate traffic Not effective in
(2017) graphs, and k- volume, with density are collected from speed and vehicle heavy traffic and
means UAV videos. The model detects the traffic count curved roads
clustering flow directions and evaluates stream
parameters separately. The model employs
KLT tracker, connected graphs, and k-means
clustering
Raj et al. KNN, ANN Traffic density is estimated using positional Lesser mean Scalability issues,
(2016) sensors that can measure parameters like absolute sensitive to noisy
volume and mean speed time. The model percentage errors data
uses KNN and ANN for estimating and (MAPEs)
predicting purposes
Chung and Deep CNN The model uses a supervised learning (deep Acccurate Cannot distinguish
Sohn (2018) convolutional neural network) for counting between moving
the number of vehicles on a strip of road and stopped
based on video imagery. The method vehicles
collectively counts the vehicles in a human-
like manner
169

Zhang et al. FCN The fully convolutional networks (FCN) Significant Lacks robustness
(2017) method used on webcam images estimates reduction of
vehicle count and density using a framework MAPEs compared
for residual learning for performing end-to- to other methods
end density forecasting, permitting random
image resolution, while adapting to varying
vehicle perspectives and scales
Zonoozi et al. Periodic- The model uses P-CRN to forecast traffic Good accuracy Not effective in
(2018) Convolutional density. P-CRN integrates with CRN to poor light
Recurrent capture temporal and spatial correlations conditions
Network (P- precisely, it performs learning and
CRN) incorporating of clear periodic
representations and optimization can be done
using multi-stage ahead forecasting
Perez-Murueta Deep learning, The vehicle redirection model for avoiding Inexpensive, easy Not scalable
et al. (2019) k Shortest Path congestion employs deep learning to forecast to implement
algorithm the state of the traffic network, and this data
identifies the areas, which could be
congested. Vehicles that are going to cross
these areas are redirected through alternate
routes created by entropy-balanced k
Shortest Path algorithm. The model uses data
collected in real time by a group of probe
cars for identifying non-recurring
congestions
Managing Networks with Computational Intelligence

Of late, Computational Intelligence (CI), which is a computational technique and algorithm
derived from nature, has been employed for managing traffic and reducing congestion. Among such
traffic management approaches, ant, bee, and genetic algorithms are widely used for Vehicle Traffic
Routing (VTR). The following subsections provide the review of the studies, which used any of these
algorithms to detect and manage traffic congestion.
Ant algorithms
Cong et al. (2013) proposed ant colony routing, a forecast-based method, which utilized colored
ants with stench pheromone for routing vehicles and reducing congestion. The stench pheromone is used
for distributing vehicles using street maps. The findings revealed effective detection of vehicles on the
streets. However, difficulties in theoretical analysis restricted the application of ant colony routing.
Jabbarpour et al. (2014) proposed the ant-guided congestion prevention algorithm that segments
maps and mean travelling speed forecasts to minimize congestion by arriving at the minimally congested
nearest paths. On simulation, the algorithm surpassed current methods in speed and mean travel duration,
and managed non-repeating congestion better. However, the variation in probability distribution with
respect to different iterations limited the detection accuracy.
Wang et al. (2015) proposed the multi-metric ant colony optimization algorithm, an ant colony
VTR system, which utilizes the least distance and time, optimal road situation, or the combined sequence,
suited to users’ preferences (pheromones), to consider real-time data for the VTR process. The algorithm
uses the TOPSIS algorithm to select the optimal route. The findings revealed that the proposed algorithm
achieved suitable routes as per the varied requirements in an easy and rapid manner.
Bee
Bee algorithms are derived from the foraging nature of bees. Artificial bees (ABs) collectively find
solutions to intricate optimization issues in bee algorithms. Like ants, two directions, i.e., forward and
reverse, are assigned to nectar-seeking ABs. Forward implies that ABs are positioned in the origin and
search their environment individually by creating many partial solutions. Reverse implies that ABs return
to their origins from their destinations and decide as a group.
Teodorović and DellOrco (2008) proposed a bee algorithm for solving the ride-matching issue,
and thus, the road congestion issue. Ride-matching assists commuters in sharing a vehicle during their
travel to destinations, and thus minimizes the number of vehicles. Consequently, money is saved and
stress and travel duration is reduced. However, slow convergence for local minima restricted the scope of
the bee algorithm.
Wedde and Senge (2007) proposed the distributed and self-organized Bee Jam algorithm for VTR.
This algorithm is a multi-bee VTR system created for reducing vehicular travel duration by avoiding
170
congestions. It gathers the latest and real-time traffic data by employing vehicle-to-infrastructure
communications. The findings revealed that Bee Jam algorithm succeeded in reducing travel time to a
great extent. However, the random initialization nature affects accuracy in the detection of vehicles.
Genetic Algorithms
Genetic algorithms are based on populations that provide possible solutions as chromosomes. All
chromosomes comprise of a group of items that provide a group of solutions for optimizing variables. The
evaluation of the suitability of every solution is carried out by an objective function. The three essential
operations of genetic algorithms are selecting, crossing over, and mutating.
Wen et al. (2011) developed a multi-objective VTR system that employed distance, cost, and time
as the objectives for locating routes and named it as the distance-Pareto genetic algorithm. Road map
hierarchy (2-level) was employed for reducing computing duration. Pareto- and crowding distance-based
objective functions were presented in this technique.
Yu and Lu (2012) developed a VTR system algorithm having mutation and crossover operators
redefined in a single mode with two new operators, namely, hyper-mutation and hyper-crossover,
developed for mode-to-mode functioning. A novel objective function operating on a p-dimensional vector
was employed to locate the optimum path. Routes representation was by changing-length chromosomes,
and representation of different kinds of transportation modes was by sub-chromosomes.
Dezani et al. (2014) proposed a new genetic-based VTR system that employs Petri-net method as
the objective function. This method permits the system to manage the entire street network in real time.
Vehicular ad-hoc network (VANET) is employed for capturing real-time data. Cagara et al. (2014)
developed another genetic algorithm for enhancing the optimum use of roads by reallocating vehicles
along alternative routes. The drawbacks of complex evolutionary operators and comparatively higher
computational costs restricted the application of genetic algorithms, even though they detected random
types of objectives and constrains in a satisfactory manner. Table 4 presents the computational
intelligence-based algorithms for mitigating congestion.
Table 4. Summary of computational intelligence based algorithms for avoiding congestion

Cong et al. The Ant Colony Routing method solves the Good
(2013) dynamic traffic routing issue. This effectiveness
algorithm employs artificial ‘ants’ for
searching the network, and the consequent
ants assignment identifies the network’s Theoretical analysis
splitting rates. The algorithm functions by not easy, Probability
network pruning and forecast control distribution varies
Jabbarpour An ACO algorithm was integrated with map Performs well in iteratively,
et al. (2014) segmenting and the mean speed of travel terms of average convergence time
forecast for deriving an enhanced travelling speed not certain
Ant Colony congestion avoiding method. Segmenting and time even
Optimization and short-term forecast were utilized for when usage is
(ACO) overcoming the dynamic nature and rapid high
variations of road traffic
Wang et al. The proposed multi-metric ACO algorithm Achieves suitable
(2015) achieves real-time variable route planning in routes as per
complex urban traffic. Attributes are varied
gathered from real traffic scenarios as the requirements
pheromone values of the ACO algorithm easily, good
that can create real-time route plans. Then, shortest route and
method for order preference by similarity to time planning
ideal solution technique is used in forked
roads for choosing the optimal route
Teodorović The algorithm provides a solution to the Effective, simple,
and Dell’Orco ride-matching issue, and, as a result, the and robust
(2008) congestion issue using BBO. Ride-matching
assists commuters in sharing a transport Convergence for
during their travel to their destinations, and local minima slow,
Bee-based thus minimize the number of vehicles initialization is
optimization having one occupant only random, parameters
Wedde and The BeeJam algorithm works on a multi- Travel times very for tuning are more
Senge (2007) agent method, based on the foraging nature much reduced
of bees. It has a dispersed vehicular–
infrastructural architecture
171

Wen and Gen An effective selection method was devised Can handle
(2008) for solving multi-objective route solving random types of
keeping driving distance, time, and cost objectives and
minimal, at the same time. A genetic constraints
algorithm termed distance Pareto GA is
proposed for solving the problem. The
solution assures rapid convergence on the
Pareto-optimal front and gives ample
attention to diversity
Yu and Lu This enhanced GA for route strategizing in Easy tuning of
(2012) multi-criteria/multi-modal environment uses parameters
High computation
mutation/crossover operators in single mode,
Genetic cost, complex
and hyper-mutation/hyper-crossover in inter-
Algorithm (GA) evolutionary
mode functioning. Vector-based assessment
operators
is used for representing multiple criteria. By
multi-objective ranking approach, the
optimum solutions are arrived at
Dezani et al. The methodology uses the Petri net analysis Fast solutions, no
(2014) as the GA’s fitness function. The output of overload of
this fitness function is the overall duration roadways
which vehicles take to arrive at their
destinations. An algorithm was developed
that works on the High-level Petri net
analyses for suggesting a vehicle the
optimal route based on travelling time
3.4. Accurate prediction of travel time

In the present times, accurate systems of travel time prediction are strongly required in public
transport. In an emergency, such systems help users to reach the nearest location, such as hospitals,
schools, and airports, in a minimal amount of time. In addition, people can select the mode of
transportation by knowing the travel time. Prediction of travel time can be modeled as historical, data-
driven, and model-based. The three different models of travel time prediction discussed in this section are
presented in Table 5. The approaches for solving travel time prediction can be broadly classified as
parametric, non-parametric, and hybrid solutions.
Approach 1: Parametric approaches
In the past, predicting the future requirements of traffic was based on analysis of time series by
observing historical data. In a time series, a set of observed readings x are captured at a particular time t.
The aim is to look for temporal patterns in historical traffic information and employ these results for
forecasts. Ahmed and Cook (1979) employed the Auto Regressive Integrated Moving Average (ARIMA)
approach for predicting short-term traffic, which generated better results over conventional double-
exponential smoothening and moving average approaches. Also, Hamed et al. (1995) conducted a time-
series study using the Box-Jenkins approach for forecasting short-term traffic flow and suggested that the
ARIMA method was forecasted traffic flow accurately for urban arterial streets. Following this, Williams
and Hoel (2003) proposed seasonal ARIMA, Van Der Voort et al. (1996) developed the Kohenen
ARIMA, and Williams (2001) proposed ARIMA combined with an descriptive variable for more accurate
prediction of traffic flow. Ghosh et al. (2009) came up with a multivariate structural time-series approach
for forecasting traffic volumes. Predicting future traffic flow based on historical data helps to synchronize
traffic signals to ensure the smooth flow of the overall traffic of the city. Liu and Wang (2010) proposed a
dynamic router travel time prediction approach, which was a model based on historical and current travel
time. The traffic pattern is analyzed by extrapolating the present real-time data through closely matched
historical patterns.
The Kalman Filtering method for time-series analysis was yet another model applied to the mobile
stochastic problems which could solve regression issues and reduce variance for arriving at optimal
results. Okutani and Stephanedes (1984), Xie et al. (2007), and Wang et al. (2008), among others used
this method for predicting traffic flow.
Approach 2: Non-parametric approaches
Random Forest Algorithm
Yu et al. (2018) proposed a technique called Random Forest based on Near Neighbor (RFNN).
RFNN method was calibrated and validated with real-world data, such as location and time, extracted
from GPS. The bus travel time between adjacent bus stops was calculated based on real-world data,
which was also used to train the RFNN model. In comparison to Linear Regression (LR), KNN, SVM and
172
Random Forest (RF), the RFNN model resulted in higher accuracy; however, it required longer
computation time. Lack of optimization due to the utilization of parallel computing technology proved to
be a drawback of RFNN method. Further, Gupta et al. (2018) proposed an ensemble learning approach to
predict taxi travel time. A lot of information was investigated and large features were extracted from the
database. Gradient boosting as well as random forest algorithm was used to train and evaluate the
performance. Data, such as taxi demand, time, trip length distribution, and most traversed areas, were the
points of interest. Statistically, it is observed that gradient boosting method provides better prediction
efficiency when compared to the random forest. Random forest response is faster and it is likely to
achieve greater accuracy with the increase in training data size. Lack of real time data restricted the scope
of outcomes.
Bayesian Algorithm approach
Development of various prediction models based on certain algorithms is also used in predicting
the travel time. Lee et al. (2008) proposed a prediction model based on Naive Bayesian Classification
(NBC). The findings revealed that the proposed prediction model yielded higher accuracy and speed
when applied to such cases and large dataset. The NBC-based model was scalable to any arbitrary travel
route. However, the data collection was not real time and hence the prediction accuracy was reduced.
Further, Hamner (2010) prepared a model, based on a context-dependent random forest method, for the
prediction of future travel time of cars in selected roads. Features representing local and aggregate traffic
flow were extracted to train the model. The collected data included the status of cars, such as the number
of stationary and moving cars in different routes. Although the algorithm performed better, it was
believed that the technique could be improved by using an advanced model of local traffic flow.
KNN, PCA, and Support Vector Algorithms
The communication of information between vehicle and infrastructure is another technique to
predict the travel time, which helps to determine the density and volume of the traffic. Jenelius and
Koutsopoulos (2017) suggested a probe data-based technique to predict urban network travel time. Future
travel time is calculated based on the current travel time and network details. This technique divides the
network into different sub-networks and applies Probabilistic Principal Component Analysis (PPCA) to
each network. In each network, average travel time is modelled as a function of day of the week, time of
the day and link. The suggested model has the ability to handle missing data and it is designed
considering several components, such as efficiency, flexibility, and robustness. Four months data was
taken from the Shenzhen Urban Transport Planning Centre, China, for conducting the research. The
results demonstrated that PPCA prediction outperformed KNN prediction with a high accuracy. Tian
(2018) proposed a model to predict travel time based on ANN and Support Vector Machine (SVM) under
the IoV environment. The results revealed that the least square SVM machine travel time predictor model
performed satisfactorily. However, it was only used for short travel time prediction with a limited sample
space. Further, Philip et al. (2018) suggested that a data-driven approach renders a better accuracy for
travel time when there is a high variability and lesser availability of traffic data. In order to predict urban
travel time, data was collected for 11 weeks, covering 2.8 km via Bluetooth sensors placed at certain
locations on an urban arterial corridor in Chennai, India. Support Vector Regression (SVR) model was
used in the data-driven approach. The results showed a better performance when compared with ANN.
However, Oh et al. (2018) pointed out certain drawbacks, such as the data-driven approach was specific
to the location and was therefore subject to change of parameters associated with the location. These
drawbacks made the data-driven approach less robust with geometric changes.
Deep Learning Algorithm
Artificial Neural Networks (ANNs): ANNs are effective methods for modeling traffic data
mathematically. They offer functionalities like learning and organizing on their own, and can recognize
patterns. They carry out non-linear approximating between outputs and inputs. As they are non-
parametric, no assumptions are made regarding data distribution. Smith et al. (1994), Elleuch et al.
(2016), and Padiath et al. (2009) proposed solutions based on ANN. Elleuch et al. (2016) proposed an
architecture of an intelligent congestion prediction approach based on data fusing and ANN. The
approach considered both, the real time unforeseen occurrences which influence traffic jams (e.g.
accidents) and the historical GPS information. Padiath et al. (2009) experimented using a historic
approach, an ANN-based approach, and a model-based approach to predict traffic density on Indian
roads. These were corroborated using data gathered using a video graphic method. A comparison with
actual data obtained by input-output analysis indicated that the approaches worked well. Jordan’s
173
sequential network is an Artificial Neural Network (ANN) technique proposed by More et al. (2016) for
predicting road traffic. Traffic volume is used as a prediction parameter. The ANN technique considers
real traffic data as well as aggregated past values for predicting the future traffic flow. The performance
of the network is analyzed through parameters, such as accuracy and speed. The study revealed that
prediction based on both real-time traffic and historical data was found to be more accurate than the
prediction based on only historical data.
Multilayer Perceptron (MLP): The MLP consists of several connected neurons which are
organized as hierarchical layers, and have an activation function of real values. MLP carries out two
important computations: forward and backward passes. In the former, a training data set is introduced in
the network and according to the model, the output error is calculated, while in the latter, the calculated
error adjusts network weights. Zargari et al. (2012) used MLP algorithm for traffic flow predictions in the
Rasht–Quazvin (Iran) highway. For training, Quasi-Newton BP method was used. Hou et al. (2015) used
Multi Layer Feed-forward network for predicting traffic flow using sensors for data acquisition in work-
zone freeways in St. Louis (US). Xu et al. (2017) employed Random Forest and MLP algorithm for
predicting traffic flow. Toll information was used as the data source in an expressway. Pamula (2018)
used a combined MLP/Autoencoder algorithm for predicting traffic flow in an urban road in Gliwice
(Germany). The Levenberg-Marquardt algorithm was used for training. Another variation of the MLP is
the Back Propagation Neural Network (BPNN). Xu et al. (2018) used BPNN technique for trip-oriented
travel time prediction for urban networks, based on historical vehicle trajectories. The results indicated
the daily and weekly variations of travel time, and the impact of climatic conditions on it. The findings
also revealed that the predictions were largely impacted by non-trivial conditions. The study pointed out
that the prediction based on available historical data was intriguing due to spatial and temporal
correlations, changes in weather conditions, and some random unknown factors, which lead to abrupt
variations in the travel time. These factors limit the application of BPNN technique in predicting the
travel time of vehicles.
Convolution Neural Network (CNN): The CNN neurons are connected to a smaller group of inputs
(compared to ANN), which reduces the total variables in the network. Further, CNNs are able to extract
significant and distinctive image features by using the convolution function. Yu et al. (2017) used CNN
as the prediction algorithm for traffic flow in a highway in Beijing. Adam optimization algorithm was
used for training. Liao et al. (2018) used CNN for predicting traffic flow in a freeway in California. The
training was carried out using Supervised Stochastic Gradient Descent algorithm. Chung et al. (2018)
used Deep CNN for predicting density of traffic and Gradient Descent Back-propagation algorithm for
training. Video detectors were used to gather data. Zhang et al. (2019) used CNN for predicting traffic
flow in a roadway in New York City.
Recurrent Neural Networks (RNNs): Typical ANNs cannot work with sequential and time series
information since they do not have a memory feature. However, RNNs can handle sequences and time
series data, along with temporal-spatial issues. An et al. (2011) used the RNN algorithm for predicting
traffic flow in an urban arterial road in Xian (China) and the least square algorithm for training.
Long Short Term Memory Networks (LSTMs): LSTM is a variant of RNN and is better than RNN
since it selectively remembers patterns for a longer duration. The network can work with time series
information in the short term and the long term. Ma et al. (2015a), Shao and Soong (2016), and Zhao et
al. (2017) used LSTM for predicting traffic flow, speed, Vehicle volume, Lane occupancy, and Average
velocity in freeways in Beijing. They used RTMS detector, video detector, and cameras, respectively, as
data sources. Kang et al. (2017), again, used LSTM for predicting traffic flow and Adam optimizer
algorithm for training. Liu et al. (2017) suggested similarities in the flow of traffic with respect to its time
series. They reported that these time series were affected by external factors, such as traffic accidents,
roadwork, fluctuations in traffic demand, and change in weather. This resulted in non-stationary
properties of the prediction algorithm. Under such conditions, the authors proposed a deep learning
approach along with time series to obtain reliable and accurate prediction. The study was conducted by
testing and using Long Short-Term Memory with Deep Neural Networks (LSTM-DNN) model. The
study pointed out that one-step prediction accuracy was found to be greater in short-term time series.
Stacked LSTM: Here, multiple LSTM layers are stacked forming a deep LSTM-driven neural
network having the following advantages: (i) it permits the model to learn features of raw temporal data
from varied facets at every time stage and (ii) model variables are distributed throughout the entire model
space without enhancing memory, thereby enabling the model to hasten convergence. Abbas et al. (2018)
employed stacked LSTM algorithm for predicting traffic flow, density, congestion, and speed using
motorway control system data in a highway near Stockholm. Zou et al. (2018) also used stacked LSTM
algorithm for predicting traffic flow in a freeway in New York City. They used GPS data from BikeNYC.
174
Gated Recurrent Unit (GRU): The LSTM network has an issue of vanishing gradient. To
overcome this issue, the GRU has been developed without a distinct memory block, thereby making it
more efficient and easy to use during training. Zhang and Kabuka (2018) used a GRU for predicting
traffic flow based on Caltrans PeMS sensor data from highways in California. Fu et al. (2016) used
LSTM and Gated Recurrent Unit algorithm for traffic flow prediction in a freeway and Back-propagation
Through Time algorithm for training.
Wavelet Neural Network (WNN): The WNN approach searches a wavelet group in feature space
in a manner that the intricate relationship inside the original signal would be represented. They were
developed as mechanisms for approximating functions which enables solving poor convergence problems
of conventional neural networks. Guorong and Yanping (2010), Gao et al. (2013), and Li et al. (2014)
used WNN algorithm for predicting traffic flow in urban areas in China.
Stacked Autoencoder (SAE): Here, multiple autoencoders are stacked in many concealed layers
above one another by employing a layer-wise learning algorithm. All autoencoders take bottleneck
activating output vectors as their inputs from the autoencoder layers below them. After the concealed
layers of stacked autoencoders are trained, a back-propagation method reduces the error function by
weights updating. Duan et al. (2016) and Yang et al. (2017) used SAE algorithm for prediction of traffic
flow in freeways. Further, a regression layer was used on top of the SAE layer.
Deep Belief Network (DBN): DBNs are comprised of at least one stacked Restricted Boltzmann
Machine (RBM) layer for the pre-training stage after which a feed-forward algorithm fine tunes it.
Concealed layers collect relevant features from the inputs which are forwarded as inputs to the subsequent
RBM layer that is stacked on it, thus creating a deterministic neural network. Huang et al. (2014) and
Soua et al. (2016) proposed a method for predicting traffic flow using DBN on freeways using Greedy
layerwise unsupervised learning and supervised back-propagation, respectively, for training. Zhang et al.
(2018) proposed a method for prediction of traffic flow using DBN and Genetic Algorithm and Fletcher–
Reeves conjugate gradient algorithm for training.
Approach 3: Hybrid approaches
As traffic flow information is non-linear and complex, it becomes difficult to predict results
accurately using only a single model. Therefore, hybrid modelling methods for improved prediction
results are necessary. These approaches predict traffic flow employing a combination of multiple
intelligent models for improved accuracy. Wu and Tan (2016) combined CNN and LSTM algorithms for
prediction of traffic flow on a freeway. Ma et al. (2015b) used a deep learning approach for a large-scale
transport network employing GPS data. RBM and RNN models driven by gradient descent optimizing
approach were used. The training data was separated into several mini samples, with all sets being
operated in parallel employing conventional gradient descent optimizing approach. Table 5 provides a
summary of the discussed methods.
Table 5. Summary of Travel Time Prediction Techniques
Prediction
Types Authors Advantages Disadvantages
Techniques
Parametric Ahmed and Cook (1979), Hamed et al. Time series -Ease of use -Models based on static
approaches (1995), Williams and Hoel (2003), Van Der -Minimal framework
Voort et al. (1996), Williams (2001), Ghosh computations -New observations not
et al. (2009) required used to update model
Okutani and Stephanedes (1984), Xie et al. Kalman Filtering Allows state -Can generate over- or
(2007), Wang et al. (2008) variables under-prediction
updating -Cannot handle
without unstable traffic
interruption conditions
Non- Yu et al. (2018), Gupta et al. (2018) Random Forest -Easy -Cannot predict beyond
parametric implementation the range in the
approaches -Training speed training data
is fast -Overfitting issues
Lee et al. (2008), Hamner (2010) Bayesian Can update -Higher computational
approach prediction when cost
new information
is available
Jenelius and Koutsopoulos (2017), Tian KNN, PCA, and -Robust to noise Spatial and temporal
(2018), Philip et al. (2018), Oh et al. (2018) Support Vector -Fast modeling cannot be
handled together
Smith et al. (1994), Elleuch et al. (2016), Deep Learning: Handles non- Handles complex
Padiath et al. (2009), More et al. (2016) ANN linear problems poorly
approximation
175
Zargari et al. (2012), Hou et al. (2015), Xu Deep Learning: Can self learn Cannot handle time-
et al. (2017), Pamula (2018) MLP series data
Yu et al. (2017), Liao et al. (2018), Chung et Deep Learning: Can recognize More training data
al. (2018), Zhang et al. (2019) CNN images well required for predicting
An et al. (2011) Deep Learning: Can handle Multiple parameters
RNN dynamic issues needed for updating
Ma et al. (2015a), Shao and Soong (2016), Deep Learning: Can selectively Cumbersome training
Zhao et al. (2017), Kang et al. (2017), Liu et LSTM remember needed
al. (2017), Abbas et al. (2018), Zou et al. patterns for
(2018) longer duration
Zhang and Kabuka (2018), Fu et al. (2016) Deep Learning: Easy to train Slow convergence and
GRU low learning efficiency
Guorong and Yanping (2010), Gao et al. Deep Learning: Can handle non- Lacks basis parameter
(2013), Li et al. (2014) WNN linear and selection
complex
highway traffic
flow data
Duan et al. (2016), Yang et al. (2017) Deep Learning: Accurate feature Time consuming
SAE extraction
Huang et al. (2014), Soua et al. (2016), Deep Learning: Robust Training is
Zhang et al. (2018) DBN classification complicated
Hybrid Ma et al. (2015b), Wu and Tan (2016) RNN and Can handle Noise (weather, state
approaches Restricted multiple of the roads) affects the
Boltzmann parameters for model
Machine updating
4. Conclusions and future work
The present paper conducted a literature survey to identify the efficiencies of key existing
Intelligent Transportation Systems (ITS) for managing traffic. The paper reviewed studies that examined
the performance of ITS based on its functional efficiencies. The review revealed that ITS used wireless
network systems to gather the data about vehicles and performed significantly well in comparison with
other data collection techniques. However, high expense of antennas limits its application. Further,
various studies with different technologies in managing traffic and indicated that the application of a
specific technique relies on the nature of location, where the traffic needs to be managed. The review of
studies pertaining to the application of different technologies in managing traffic revealed that the
application of deep learning technologies and certain optimization algorithms like Ant and Bee algorithms
provided better results in managing traffic. The review of various technologies used for predicting
accurate travel time through the detection of traffic congestion signposted that technologies designed on
two models provide a better accuracy in the detection of vehicular congestions, thereby predicting travel
time with a high degree of accuracy.
Thus, in conclusion, this survey on using ITS for reducing traffic congestion indicates that ITS
options are vast, involve several technologies, and have the potential to mitigate most of the traffic
problems. However, the application should rely on the nature of location, where the traffic needs to be
managed. If appropriate ITS is used according to the nature of location, then travel time and queues could
be reduced, resulting in the provision of better safety and productivity. In addition to these findings, the
review pointed out certain unexplored areas, which open the door for future researchers. These future
directions are described in the subsequent subsection.
Future directions
 Improvement in data collection

Owing to the constant mobility of vehicles, the collected data might often not be accurate,
complete, or dependable. Research on the techniques that yield better quality of data would be significant.
In the era of Internet of Things, many new upcoming sensing methods could help in the improvement of
data acquisition and quality. Additionally, research on the adoption of automatic data-capturing
techniques that reduce manual data entry can benefit data quality improvement.
 Integration of non-homogeneous data sources
Lately, there has been a proliferation of devices, each with its own standards. Traffic management
systems need to work with all such devices. Therefore, developing common standards is a challenge,
which needs a considerable collaborative research.
176
 Managing huge amounts of data

Data is reaching massive proportions with the growth of Internet of Things. The seamless
integration and processing of data is another big challenge, and much research is needed for managing the
data. The use of technologies, such as cloud computing and edge computing, could be considered. Using
Big Data technology to gather traffic data of an entire city, practical data can be provided for guiding
traffic and urban planning, which could solve problems related to traffic congestion analysis, traffic flow
prediction, and wise planning of transportation infrastructure (Wang et al., 2018).
 Privacy concerns
Huge amount of data stored via latest technologies could contain the private data of the respective
owners. The transmission of such sensitive information could be intercepted by hackers. Therefore, the
ensuring of safe traffic management systems is a challenge that requires further research.
 Detection of traffic hazards
Even though the techniques for identification of traffic hazards have been studied, they are not
efficient enough. Potential hazards must be correctly identified and transmitted to the drivers for them to
take appropriate action. The area of detecting traffic hazards still needs to be researched.
 Policy formulations
Parameters like traffic violations and accidents need to be integrated with traffic managing systems
in the future so that policy makers can use such data while framing traffic rules. Research on how to
integrate such data will prevent avoidable mishaps.
References
1. Abbas, Z., Al-Shishtawy, A., Girdzijauskas, S., Vlassov, V. (2018) Short-term traffic prediction
using long short-term memory neural networks. In: 2018 IEEE International Congress on Big Data
(BigData Congress), 57-65.
2. Ahmad, F., Basit, A., Ahmad, H., Mahmud, S. A., Khan, G. M., & Yousaf, F. Z. (2013) Feasibility
of deploying wireless sensor based road side solutions for intelligent transportation systems.
In: 2013 International Conference on Connected Vehicles and Expo (ICCVE), 320-326.
3. Ahmed, M.S., Cook, A.R. (1979) Analysis of freeway traffic time-series data by using Box-Jenkins
techniques, No. 722.
4. An, Y., Song, Q. and Zhao, X. (2011) Short-term traffic flow forecasting via echo state neural
networks. In 2011 Seventh International Conference on Natural Computation, 2, 844-847.
5. Araújo, G. B., Queiroz, M. M., de LP Duarte-Figueiredo, F., Tostes, A. I., & Loureiro, A. A. (2014).
Cartim: A proposal toward identification and minimization of vehicular traffic congestion for vanet.
In: 2014 IEEE symposium on computers and communications (ISCC), 1-6.
6. Avatefipour, O., Sadry, F. (2018) Traffic management system using IoT technology-A comparative
review. In: 2018 IEEE International Conference on Electro/Information Technology (EIT),
1041-1047.
7. Barros, J., Araujo, M., Rossetti, R. J. (2015) Short-term real-time traffic prediction methods: A
survey. In: 2015 International Conference on Models and Technologies for Intelligent
Transportation Systems (MT-ITS), 132-139.
8. Bell, M. G., Trozzi, V., Hosseinloo, S. H., Gentile, G., & Fonzone, A. (2012) Time-dependent
Hyperstar algorithm for robust vehicle navigation. Transportation Research Part A: Policy and
Practice, 46(5), 790-800.
9. Bernas, M., Płaczek, B., & Korski, W. (2018) Wireless Network with Bluetooth Low Energy
Beacons for Vehicle Detection and Classification. In: International Conference on Computer
Networks, 429-444. Springer, Cham.
10. Bruno, R., & Nurchis, M. (2013) Robust and efficient data collection schemes for vehicular
multimedia sensor networks. In: 2013 IEEE 14th International Symposium on" A World of Wireless,
Mobile and Multimedia Networks" (WoWMoM), 1-10. IEEE.
11. Cagara, D., Bazzan, A. L., & Scheuermann, B. (2014) Getting you faster to work: A genetic
algorithm approach to the traffic assignment problem. In: Proceedings of the Companion
Publication of the 2014 Annual Conference on Genetic and Evolutionary Computation, 105-106.
ACM.
177
12. Chao, K. H., & Chen, P. Y. (2014) An intelligent traffic flow control system based on radio
frequency identification and wireless sensor networks. International journal of distributed sensor
networks, 10(5), 694545.
13. Chen, C. P., Zhou, J., & Zhao, W. (2012) A real-time vehicle navigation algorithm in sensor
network environments. IEEE Transactions on Intelligent Transportation Systems, 13(4), 1657-1666.
14. Chen, X., Chen, R. (2019) A Review on Traffic Prediction Methods for Intelligent Transportation
System in Smart Cities. In: 2019 12th International Congress on Image and Signal Processing,
BioMedical Engineering and Informatics (CISP-BMEI), 1-5. IEEE.
15. Chhatpar, P., Doolani, N., Shahani, S., Priya, R.L. (2018) Machine learning solutions to vehicular
traffic congestion. In: 2018 International Conference on Smart City and Emerging Technology
(ICSCET), 1-4. IEEE.
16. Choi, O., Kim, S., Jeong, J., Lee, H. W., & Chong, S. (2015) Delay-optimal data forwarding in
vehicular sensor networks. IEEE transactions on vehicular technology, 65(8), 6389-6402.
17. Chowdhury, M., Hasan, M., Safait, S. (2018) A novel approach to forecast traffic congestion using
CMTF and machine learning (Doctoral dissertation, BRAC University).
18. Chung, J., & Sohn, K. (2018) Image-Based Learning to Measure Traffic Density Using a Deep
Convolutional Neural Network. IEEE Transactions on Intelligent Transportation Systems, 19(5),
1670–1675.
19. Collotta, M., Pau, G., Scatà, G., & Campisi, T. (2014) A dynamic traffic light management system
based on wireless sensor networks for the reduction of the red-light running phenomenon. Transport
and Telecommunication, 15(1), 1-11.
20. Cong, Z., De Schutter, B., & Babuška, R. (2013) Ant colony routing algorithm for freeway
networks. Transportation Research Part C: Emerging Technologies, 37, 1-19.
21. Cordeschi, N., Amendola, D., Shojafar, M., Naranjo, P. G. V., & Baccarelli, E. (2015). Memory and
memoryless optimal time-window controllers for secondary users in vehicular networks.
In: Proceedings of the International Symposium on Performance Evaluation of Computer and
Telecommunication Systems, 1-7. Society for Computer Simulation International.
22. de Souza, A. M., Yokoyama, R. S., Botega, L. C., Meneguette, R. I., & Villas, L. A. (2015)
Scorpion: A solution using cooperative rerouting to prevent congestion and improve traffic
condition. In: 2015 IEEE International Conference on Computer and Information Technology;
Ubiquitous Computing and Communications; Dependable, Autonomic and Secure Computing;
Pervasive Intelligence and Computing, 497-503. IEEE.
23. de Souza, A. M., Yokoyama, R. S., Maia, G., Loureiro, A., & Villas, L. (2016) Real-time path
planning to prevent traffic jam through an intelligent transportation system. In: 2016 IEEE
Symposium on Computers and Communication (ISCC), 726-731. IEEE.
24. de Souza, A. M., Brennand, C. A., Yokoyama, R. S., Donato, E. A., Madeira, E. R., Villas, L. A.
(2017) Traffic management systems: A classification, review, challenges, and future
perspectives. International Journal of Distributed Sensor Networks, 13(4), p.1550147716683612.
25. Dezani, H., Bassi, R. D., Marranghello, N., Gomes, L., Damiani, F., & Da Silva, I. N. (2014)
Optimizing urban traffic flow using Genetic Algorithm with Petri net analysis as fitness
function. Neurocomputing, 124, 162-167.
26. Duan, Y., Lv, Y. and Wang, F.Y. (2016) Performance evaluation of the deep learning approach for
traffic flow prediction at different times. In: 2016 IEEE International Conference on Service
Operations and Logistics, and Informatics (SOLI), 223-227. IEEE.
27. Elleuch, W., Wali, A. and Alimi, A.M. (2016) Intelligent Traffic Congestion Prediction System
Based on ANN and Decision Tree Using Big GPS Traces. In: International conference on intelligent
systems design and applications, 478-487. Springer, Cham.
28. Emo, A. K., Matthews, G., & Funke, G. J. (2016) The slow and the furious: Anger, stress and risky
passing in simulated traffic congestion. Transportation research part F: traffic psychology and
behaviour, 42, 1-14.
29. Fan, S. K. S., Su, C. J., Nien, H. T., Tsai, P. F., & Cheng, C. Y. (2018) Using machine learning and
big data approaches to predict travel time based on historical and real-time data from Taiwan
electronic toll collection. Soft Computing, 22(17), 5707-5718.
30. Faye, S., Chaudet, C., & Demeure, I. (2012) A distributed algorithm for adaptive traffic lights
control. In: 2012 15th International IEEE Conference on Intelligent Transportation Systems,
1572-1577. IEEE.
31. Friesen, M., Jacob, R., Grestoni, P., Mailey, T., Friesen, M. R., & McLeod, R. D. (2014) Vehicular
traffic monitoring using bluetooth scanning over a wireless sensor network. Canadian Journal of
Electrical and Computer Engineering, 37(3), 135-144.
178
32. Fu, R., Zhang, Z. and Li, L. (2016) November. Using LSTM and GRU neural network methods for
traffic flow prediction. In: 2016 31st Youth Academic Annual Conference of Chinese Association of
Automation (YAC), 324-328. IEEE.
33. Gao, J., Leng, Z., Qin, Y., Ma, Z., Liu, X. (2013) Short-term traffic flow forecasting model based on
wavelet neural network. In: 2013 25th Chinese control and decision conference (CCDC),
5081-5084. IEEE.
34. Ghosh, B., Basu, B., O'Mahony, M. (2009) Multivariate short-term traffic flow forecasting using
time-series analysis. IEEE transactions on intelligent transportation systems, 10(2), 246-254.
35. Ghosh, R.; Akula, A.; Kumar, S.; Sardana, H. (2015) Time–frequency analysis based robust vehicle
detection using seismic sensor. J. Sound Vibr., 346, 424–434.
36. Guo, F., Polak, J.W., Krishnan, R. (2010) Comparison of modelling approaches for short term traffic
prediction under normal and abnormal conditions. In: 13th International IEEE Conference on
Intelligent Transportation Systems, 1209-1214. IEEE.
37. Guorong, G., Yanping, L. (2010) Traffic flow forecasting based on PCA and wavelet neural
network. In: 2010 International Conference of Information Science and Management Engineering,
1, 158-161. IEEE.
38. Gupta, B., Awasthi, S., Gupta, R., Ram, L., Kumar, P., Prasad, B. R., & Agarwal, S. (2018) Taxi
travel time prediction using ensemble-based random forest and gradient boosting model.
In: Advances in Big Data and Cloud Computing, 63-78. Springer, Singapore.
39. Haferkamp, M., Al-Askary, M., Dorn, D., Sliwa, B., Habel, L., Schreckenberg, M., & Wietfeld, C.
(2017) Radio-based traffic flow detection and vehicle classification for future smart cities. In: 2017
IEEE 85th Vehicular Technology Conference (VTC Spring), 1-5. IEEE.
40. Hamed, M.M., Al-Masaeid, H.R., Said, Z.M.B. (1995) Short-term prediction of traffic volume in
urban arterials. Journal of Transportation Engineering, 121(3), 249-254.
41. Hamner, B. (2010) Predicting travel times with context-dependent random forests by modeling local
and aggregate traffic flow. In: 2010 IEEE International Conference on Data Mining Workshops,
1357-1359. IEEE.
42. Hong, W.C. (2011) Traffic flow forecasting by seasonal SVR with chaotic simulated annealing
algorithm. Neurocomputing, 74(12-13), 2096-2107.
43. Hou, Y., Edara, P., Sun, C. (2014) Traffic flow forecasting for urban work zones. IEEE transactions
on intelligent transportation systems, 16(4), 1761-1770.
44. Huang, W., Song, G., Hong, H., Xie, K. (2014) Deep architecture for traffic flow prediction: deep
belief networks with multitask learning. IEEE Transactions on Intelligent Transportation
Systems, 15(5), 2191-2201.
45. Jabbarpour, M. R., Jalooli, A., Shaghaghi, E., Noor, R. M., Rothkrantz, L., Khokhar, R. H., &
Anuar, N. B. (2014) Ant-based vehicle congestion avoidance system using vehicular networks.
Engineering Applications of Artificial Intelligence, 36, 303-319.
46. Jenelius, E., & Koutsopoulos, H. N. (2017) Urban network travel time prediction based on a
probabilistic principal component analysis model of probe data. IEEE Transactions on Intelligent
Transportation Systems, 19(2), 436-445.
47. Kang, D., Lv, Y., Chen, Y.Y. (2017) Short-term traffic flow prediction with LSTM recurrent neural
network. In: 2017 IEEE 20th International Conference on Intelligent Transportation Systems
(ITSC), 1-6. IEEE.
48. Ke, R., Li, Z., Kim, S., Ash, J., Cui, Z., & Wang, Y. (2017) Real-Time Bidirectional Traffic Flow
Parameter Estimation From Aerial Videos. IEEE Transactions on Intelligent Transportation
Systems, 18(4), 890–901.
49. Khan, N. A. (2017) Real time predictive monitoring system for urban transport (Doctoral
dissertation, Kingston University).
50. Lee, H., Chowdhury, N. K., & Chang, J. (2008) A new travel time prediction method for intelligent
transportation systems. In: International Conference on Knowledge-Based and Intelligent
Information and Engineering Systems, 473-483. Springer, Berlin, Heidelberg.
51. Li, D.M., Liu, B.I.N. (2014) Modeling and prediction of highway traffic flow based on wavelet
neural network. In: 2014 International Conference on Machine Learning and Cybernetics, 2,
675-679. IEEE.
52. Liang, Z., Wakahara, Y. (2013) City traffic prediction based on real-time traffic information for
intelligent transport systems. In: 2013 13th International Conference on ITS Telecommunications
(ITST), 378-383. IEEE.
179
53. Liao, S., Chen, J., Hou, J., Xiong, Q., Wen, J. (2018) Deep convolutional neural networks with
random subspace learning for short-term traffic flow prediction with incomplete data. In: 2018
International Joint Conference on Neural Networks (IJCNN), 1-6. IEEE.
54. Liu, W., & Wang, Z. (2010) Dynamic router real-time travel time prediction based on a road
network. In: International Symposium on Information and Automation, 723-729. Springer, Berlin,
Heidelberg.
55. Liu, Y., & Li, S. (2011) A vehicle-logo location approach based on edge detection and projection.
In: Proceedings of 2011 IEEE International Conference on Vehicular Electronics and Safety,
165-168. IEEE.
56. Liu, Y., Wang, Y., Yang, X., & Zhang, L. (2017). Short-term travel time prediction by deep
learning: A comparison of different LSTM-DNN models. In: 2017 IEEE 20th International
Conference on Intelligent Transportation Systems (ITSC), 1-8. IEEE.
57. Ma, X., Dai, Z., He, Z., Ma, J., Wang, Y., Wang, Y. (2017) Learning traffic as images: a deep
convolutional neural network for large-scale transportation network speed
prediction. Sensors, 17(4), 818.
58. Ma, X., Tao, Z., Wang, Y., Yu, H., Wang, Y. (2015a) Long short-term memory neural network for
traffic speed prediction using remote microwave sensor data. Transportation Research Part C:
Emerging Technologies, 54, 187-197.
59. Ma, X., Yu, H., Wang, Y., Wang, Y. (2015b) Large-scale transportation network congestion
evolution prediction using deep learning theory. PloS one, 10(3), p.e0119044.
60. Meneguette, R. I., Geraldo Filho, P. R., Bittencourt, L. F., Ueyama, J., Krishnamachari, B., & Villas,
L. A. (2015) Enhancing intelligence in inter-vehicle communications to detect and reduce
congestion in urban centers. In: 2015 IEEE Symposium on Computers and Communication (ISCC),
1-6. IEEE.
61. Min, W., & Wynter, L. (2011) Real-time road traffic prediction with spatio-temporal
correlations. Transportation Research Part C: Emerging Technologies, 19(4), 606-616.
62. Mohan, P., Padmanabhan, V.N., & Ramjee, R. (2008) Nericell: rich monitoring of road and traffic
conditions using mobile smartphones. In: Proceedings of the 6th ACM conference on Embedded
network sensor systems, 323-336. ACM.
63. More, R., Mugal, A., Rajgure, S., Adhao, R.B., & Pachghare, V.K. (2016) Road traffic prediction
and congestion control using Artificial Neural Networks. In: 2016 International Conference on
Computing, Analytics and Security Trends (CAST), 52-57. IEEE.
64. MTTM Movement-Travel Times Methodology (2018) URL:
https://d3i4yxtzktqr9n.cloudfront.net/web-movement/static/pdfs/Movement-
TravelTimesMethodology-76002ded22.pdf
65. Na, Y., Guo, Y., Fu, Q., & Yan, Y. (2015) An acoustic traffic monitoring system: Design and
implementation. In: 2015 IEEE 12th Intl Conf on Ubiquitous Intelligence and Computing and 2015
IEEE 12th Intl Conf on Autonomic and Trusted Computing and 2015 IEEE 15th Intl Conf on
Scalable Computing and Communications and Its Associated Workshops (UIC-ATC-ScalCom),
119-126. IEEE.
66. Oh, S., Byon, Y.J., Jang, K., & Yeo, H. (2018) Short-term travel-time prediction on highway: A
review on model-based approach. KSCE Journal of Civil Engineering, 22(1), 298-310.
67. Okutani, I., Stephanedes, Y.J. (1984) Dynamic prediction of traffic volume through Kalman filtering
theory. Transportation Research Part B: Methodological, 18(1), 1-11.
68. Padiath, A., Vanajakshi, L., Subramanian, S.C., Manda, H. (2009) October. Prediction of traffic
density for congestion analysis under Indian traffic conditions. In: 2009 12th International IEEE
Conference on Intelligent Transportation Systems, 1-6). IEEE.
69. Pamuła, T. (2018) Impact of data loss for prediction of traffic flow on an urban road using neural
networks. IEEE Transactions on Intelligent Transportation Systems, 20(3), 1000-1009.
70. Papageorgiou, M., Diakaki, C., Dinopoulou, V., Kotsialos, A., Wang, Y. (2003) Review of road
traffic control strategies. In: Proceedings of the IEEE, 91(12), 2043-2067.
71. Perez-Murueta, P., Gómez-Espinosa, A., Cardenas, C., & Gonzalez-Mendoza, M. (2019) Deep
Learning System for Vehicular Re-Routing and Congestion Avoidance. Applied Sciences, 9(13),
2717.
72. Pfleiderer, R.H., & Dieterich, M. (1995) New roads generate new traffic. World Transport Policy
and Practice, 1(1), 29-31.
73. Philip, A. M., Ramadurai, G., & Vanajakshi, L. (2018) Urban Arterial Travel Time Prediction Using
Support Vector Regression. Transportation in Developing Economies, 4(1), 7.
180
74. Polson, N. G., Sokolov, V. O. (2017) Deep learning for short-term traffic flow prediction.
Transportation Research Part C: Emerging Technologies, 79, 1-17.
75. Raj, J., Bahuleyan, H., Vanajakshi, L. D. (2016) Application of data mining techniques for traffic
density estimation and prediction. Transportation Research Procedia, 17, 321-330.
76. Rivas, J., Wunderlich, R., & Heinen, S. J. (2016) Road vibrations as a source to detect the presence
and speed of vehicles. IEEE Sensors Journal, 17(2), 377-385.
77. Saqib, M., Lee, C. (2010) Traffic control system using wireless sensor network. In: 2010 The 12th
International Conference on Advanced Communication Technology (ICACT), 1, 352-357. IEEE.
78. Sen, R., Sevani, V., Sharma, P., Koradia, Z., & Raman, B. (2009) Challenges in communication
assisted road transportation systems for developing regions. NSDR’09.
79. Shao, H., Soong, B.H. (2016) Traffic flow prediction with long short-term memory networks
(LSTMs). In: 2016 IEEE Region 10 Conference (TENCON), 2986-2989. IEEE.
80. Shen, W., Wynter, L. (2012) Real-time road traffic fusion and prediction with GPS and fixed-sensor
data. In: 2012 15th International Conference on Information Fusion, 1468-1475. IEEE.
81. Shojafar, M., Cordeschi, N., Amendola, D., & Baccarelli, E. (2015) Energy-saving adaptive
computing and traffic engineering for real-time-service data centers. In: 2015 IEEE International
Conference on Communication Workshop (ICCW), 1800-1806. IEEE.
82. Singh, G., Bansal, D., & Sofat, S. (2014) Intelligent Transportation System for Developing
Countries-A Survey. International Journal of Computer Applications, 85(3).
83. Sładkowski, A., & Pamuła, W. eds. (2016) Intelligent transportation systems-problems and
perspectives, Vol. 303. Springer International Publishing.
84. Smith, B .L., Demetsky, M. J. (1994) Short-term traffic flow prediction models-a comparison of
neural network and nonparametric regression approaches. In: Proceedings of IEEE International
Conference on Systems, Man and Cybernetics, 2, 1706-1709. IEEE.
85. Soua, R., Koesdwiady, A., Karray, F. (2016) Big-data-generated traffic flow prediction using deep
learning and dempster-shafer theory. In: 2016 International joint conference on neural networks
(IJCNN), 3195-3202. IEEE.
86. Teodorović, D., & Dell’Orco, M. (2008) Mitigating traffic congestion: solving the ride-matching
problem by bee colony optimization. Transportation Planning and Technology, 31(2), 135-152.
87. Tian, X. (2018) Research on Travel Time Prediction under Internet of Vehicles. In: 2018
International Conference on Intelligent Transportation, Big Data & Smart City (ICITBS), 38-40.
IEEE.
88. Underwood, R.T. (1990) Traffic management: an introduction.
89. Van Arem, B., Kirby, H.R., Van Der Vlist, M.J., Whittaker, J.C. (1997) Recent advances and
applications in the field of short-term traffic forecasting. International journal of forecasting, 13(1),
1-12.
90. Van Der Voort, M., Dougherty, M., Watson, S. (1996) Combining Kohonen maps with ARIMA
time series models to forecast traffic flow. Transportation Research Part C: Emerging
Technologies, 4(5), 307-318.
91. Wang Z., Li J., Fang M., & Li, Y. (2015) A multimetric ant colony optimization algorithm for
dynamic path planning in vehicular networks. Int J Distrib Sens Netw, 11(10).
92. Wang, Y., Papageorgiou, M., Messmer, A. (2008) Real-time freeway traffic state estimation based
on extended Kalman filter: Adaptive capabilities and real data testing. Transportation Research Part
A: Policy and Practice, 42(10), 1340-1358.
93. Wang, Z., Yu, Y., Ju, D. (2018) Analysis and Prediction of Urban Traffic Congestion Based on Big
Data. International Journal on Data Science and Technology, 4(3), p.100.
94. Wedde, H.F., & Senge, S. (2013) BeeJamA: A distributed, self-adaptive vehicle routing guidance
approach. IEEE Transactions on Intelligent Transportation Systems, 14(4), 1882-1895.
95. Wen, F., Gen, M., & Yu, X. (2011) A new multiobjective genetic algorithm for route selection. IEEJ
Transactions on Electronics, Information and Systems, 131(3), 619-625.
96. Williams, B. M., Hoel, L. A. (2003) Modeling and forecasting vehicular traffic flow as a seasonal
ARIMA process: Theoretical basis and empirical results. Journal of transportation
engineering, 129(6), 664-672.
97. Williams, B.M. (2001) Multivariate vehicular traffic flow prediction: evaluation of ARIMAX
modeling. Transportation Research Record, 1776(1), 194-200.
98. Won, M., Zhang, S., & Son, S. H. (2017) WiTraffic: Low-cost and non-intrusive traffic monitoring
system using WiFi. In: 2017 26th International Conference on Computer Communication and
Networks (ICCCN), 1-9. IEEE.
181
99. Wu, Y., Tan, H. (2016) Short-term traffic flow forecasting with spatial-temporal correlation in a
hybrid deep learning framework. arXiv preprint arXiv:1612.01022.
100. Xie, Y., Zhang, Y., Ye, Z. (2007) Short‐term traffic volume forecasting using Kalman filter with
discrete wavelet decomposition. Computer‐Aided Civil and Infrastructure Engineering, 22(5),
326-334.
101. Xu, D., Shi, Y. (2017) A combined model of random forest and multilayer perceptron to forecast
expressway traffic flow. In: 2017 7th IEEE International Conference on Electronics Information
and Emergency Communication (ICEIEC), 448-451. IEEE.
102. Xu, T., Li, X., & Claramunt, C. (2018) Trip-oriented travel time prediction (TOTTP) with historical
vehicle trajectories. Frontiers of earth science, 12(2), 253-263.
103. Yang, H.F., Dillon, T.S., Chen, Y.P.P. (2016) Optimized structure of the traffic flow forecasting
model with a deep learning approach. IEEE transactions on neural networks and learning
systems, 28(10), 2371-2381.
104. Yu, B., Wang, H., Shan, W., & Yao, B. (2018) Prediction of bus travel time using random forests
based on near neighbors. Computer Aided Civil and Infrastructure Engineering, 33(4), 333-350.
105. Yu, B., Yin, H., Zhu, Z. (2017) Spatio-temporal graph convolutional networks: A deep learning
framework for traffic forecasting. arXiv preprint arXiv:1709.04875.
106. Yu, H., & Lu, F. (2012) A multi-modal route planning approach with an improved genetic
algorithm. Advances in Geo-Spatial Information Science, 193.
107. Zargari, S. A., Siabil, S. Z., Alavi, A. H., Gandomi, A. H. (2012) A computational intelligence‐based
approach for short‐term traffic flow prediction. Expert Systems, 29(2), 124-142.
108. Zhang, D. and Kabuka, M.R., 2018. Combining weather condition data to predict traffic flow: a
GRU-based deep learning approach. IET Intelligent Transport Systems, 12(7), pp.578-585.
109. Zhang, J., Zheng, Y., Sun, J., Qi, D. (2019) Flow prediction in spatio-temporal networks based on
multitask deep learning. IEEE Transactions on Knowledge and Data Engineering, 32(3), 468-478.
110. Zhang, S., Wu, G., Costeira, J. P., & Moura, J. M. (2017) Fcn-rlstm: Deep spatio-temporal neural
networks for vehicle counting in city cameras. In: Proceedings of the IEEE international conference
on computer vision, 3667-3676.
111. Zhang, Y., Huang, G. (2018) Traffic flow prediction model based on deep belief network and
genetic algorithm. IET Intelligent Transport Systems, 12(6), 533-541.
112. Zhao, J., Gao, Y., Tang, J., Zhu, L., & Ma, J. (2018) Highway Travel Time Prediction Using Sparse
Tensor Completion Tactics and-Nearest Neighbor Pattern Matching Method. Journal of Advanced
Transportation.
113. Zhao, Z., Chen, W., Wu, X., Chen, P.C., Liu, J. (2017) LSTM network: a deep learning approach for
short-term traffic forecast. IET Intelligent Transport Systems, 11(2), 68-75.
114. Zhou, J., Chen, C.P., Chen, L., & Zhao, W. (2013) A user-customizable urban traffic information
collection method based on wireless sensor networks. IEEE Transactions on Intelligent
Transportation Systems, 14(3), 1119-1128.
115. Zonoozi, A., Kim, J. J., Li, X. L., & Cong, G. (2018) Periodic-CRN: A Convolutional Recurrent
Model for Crowd Density Prediction with Recurring Periodic Patterns. In: IJCAI, 3732-3738.
116. Zou, Z., Gao, P., Yao, C. (2018) City-Level Traffic Flow Prediction via LSTM Networks.
In: Proceedings of the 2nd International Conference on Advances in Image Processing, 149-153.
182
© 2021. This work is published under
http://creativecommons.org/licenses/by-nc-nd/3.0 (the “License”).
Notwithstanding the ProQuest Terms and Conditions, you may use this
content in accordance with the terms of the License.
View publication stats

Intelligent Traffic Management: A Review of Challenges, Solutions, and Future Perspectives

Uploaded by

Copyright:

Available Formats

Intelligent Traffic Management: A Review of Challenges, Solutions, and Future Perspectives

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Intelligent Traffic Management: A Review of Challenges, Solutions, and Future Perspectives

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Intelligent Trafﬁc Management: A Review of Challenges, Solutions, and

Article in Transport and Telecommunication · April 2021

Intelligent Transportation System View project

Software Implementation of an Automatic Movable Road Barrier View project

The user has requested enhancement of the downloaded file.

Transport and Telecommunication, 2021, volume 22, no. 2, 163–182

INTELLIGENT TRAFFIC MANAGEMENT: A REVIEW OF

3. Effective Traffic Control and Management using ITS

Figure 1. Functional categories for congestion management solutions by using ITS

3.1. Traffic Data with Travel Time Constraints

Reference Technology Principles Advantages Disadvantages

3.2. Traffic Management Systems

Authors Technology Principle Advantages Disadvantages

Authors Technology Principle Advantages Disadvantages

3.3. Congestion Avoidance Schemes

Table 3. Summary of machine learning approaches to mitigate congestion

Reference Algorithm Principle Advantages Disadvantages

Reference Algorithm Principle Advantages Disadvantages

Managing Networks with Computational Intelligence

Reference Algorithm Principle Advantages Disadvantages

Reference Algorithm Principle Advantages Disadvantages

3.4. Accurate prediction of travel time

4. Conclusions and future work

 Improvement in data collection

 Managing huge amounts of data

View publication stats

You might also like