ISSN: 2320-5407 Int. J. Adv. Res.

11(11), 64-74

Journal Homepage: -www.journalijar.com

Article DOI:10.21474/IJAR01/17802
DOI URL: http://dx.doi.org/10.21474/IJAR01/17802

RESEARCH ARTICLE
AN AGENT BASED MODEL TO IMPROVE TRAFFIC FLOW AND ASSESS THE IMPACT OF SMART
TRAFFIC LIGHT

N’golo Konate1, Mory Richard Batiebo2, Mamadou Diarra1 and Agboka Komi Mensah3
1. Université Felix Houphouët Boigny(UFHB).
2. Université Virtuelle de Côte d'Ivoire (UVCI).
3. International Centre of Insect Physiology and Ecology (ICIPE).
……………………………………………………………………………………………………....
Manuscript Info Abstract
……………………. ………………………………………………………………
Manuscript History The Traffic in developing country undergoes severs perturbations due
Received: 05 September 2023 to either a misunderstanding of the physical phenomenon or in bad
Final Accepted: 09 October 2023 calibration of equipment’s such as traffic light. Due to this high
Published: November 2023 complexity, mathematical models fail to well represent different
scenario. Those issues are mostly due to the high amount of parameter
Key words:-
Agent-Based Model, Dynamic Traffic which should be taken into consideration when dealing with some
Light, Urban Mobility, Smart Mobility, mathematical model and the drivers’ behaviors. Since agent-based
Urban Environment, Intelligent Road model is a computational simulation that represents individual entities
Traffic, Urban Sustainability
(called agents) and their interactions within a defined environment. It is
a powerful tools used to investigate human behavior impact in a
mathematical model. Therefore, this paper proposed a novel agent-
based approach framework for effectively making traffic light timing
dynamic according to traffic density. Moreover, based on different
rules such as lane changing, stop and goat the traffic light, we develop
an Agent based model to represent the relationships between the traffic
changing and its environment. The use of agent-based models for
traffic light time optimization offers a more granular, adaptive, and
integrative approach that better capture the complexities of real-world
traffic scenarios. Finally, the experimental result is applied on synthetic
data and compared to other results in almost the same context for both
static and dynamic traffic light. The results show that the travel time
distribution is less for dynamic traffic light.

Copy Right, IJAR, 2023,. All rights reserved.

……………………………………………………………………………………………………....
Introduction:-
Traffic congestion has become a pervasive problem in urban areas across the globe, giving rise to substantial
economic, environmental, and social consequences (Huang Z. a., 2023) (Litman, 2021).As urban populations
continue to expand, and cities experience unprecedented growth, the need for efficient and sustainable transportation
systems has never been more crucial. Researchers and policymakers have been exploring various approaches to
address these challenges, including the development of more accurate traffic flow simulation models, the
optimization of traffic light control systems, and the integration of intelligent transportation systems (ITS)
(Barmpounakis, Intelligent transportation systems and powered two wheelers traffic, 2015) (Zhu, 2019)

Corresponding Author:- Mory Richard Batiebo

Address:- Université Virtuelle de Côte d'Ivoire (UVCI). 64
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

One of the primary factors contributing to traffic congestion is vehicle density, defined as the number of vehicles per
unit of road space (Berhanu, 2023). Higher vehicle densities often result in increased travel times, fuel consumption,
and greenhouse gas emissions, while simultaneously decreasing road safety and overall quality of life for residents
(Kawajiri, 2020). Consequently, a comprehensive understanding of the relationship between vehicle density and
traffic flow is essential for the development of effective traffic management strategies that can mitigate the negative
effects of congestion.

This paper seeks to contribute to the existing body of knowledge by presenting a comprehensive multi-lane traffic
flow simulation model that investigates the impact of lanes number and traffic light timings on urban traffic
congestion. Our model employs a simple grid-based representationof a road network with multiple lanes and
integrates traffic light control mechanisms that can be adjusted to study their effects on traffic flow. We further
examine the potential of incorporating intelligent transportation systems (ITS) and machine learning techniques to
optimize traffic light control strategies based on real-time data (Barmpounakis, Unmanned Aerial Aircraft Systems
for transportation engineering: Current practice and future challenges, 2016).

By analyzing simulation’s results under various scenarios and traffic conditions, we aim to provide a deeper
understanding of the factors that contribute to traffic congestion and explore potential strategies for mitigating its
adverse effects. Moreover, our study seeks to inform policymakers and urban planners about the potential benefits
and challenges associated with implementing advanced traffic management systems and inspire further research in
this crucial field.

The aim of this paper is to give the traffic engineer a simulation perspective of traffic evaluation in different
scenarios.

RelatedWork:
Traffic light is a crucial infrastructure in mobility management (Nedim, 2023). However, it can be a source of traffic
congestion if it’s not well calibrated. Therefore, the researchers have evaluated the efficiency through many
contexts (Chen, 2018) (Nagel, 1992) (Nedim, 2023).

In analyzing the complex dynamic behavior of traffic streams, simulation tools have a prominent place (Qadri,
2020).

Traffic Control Signal strategies have been classified in fixed-time, adaptive, and actuated control strategies
(Ribeiro, 2016). If fixed time is appropriate for traffic signal when the flow is stable and constant, it becomes
inadequate when the traffic density becomes heavy. Conversely to fixed-time control strategy, researchers have
developed and adaptive control strategy using mathematical model (Bae, 2019).

The duration of green and red light phases significantly impacts the efficiency of traffic movement, and optimizing
these timings can help alleviate traffic issues in urban areas (Huang H. a., 2014). Nonetheless, determining the
optimal traffic light timings is a complex problem contingent on numerous factors, such as traffic demand, road
network characteristics, and traffic flow dynamics (Nigam, 2023).

Over the years, researchers have employed a variety of computational methods and simulation models to study
traffic flow and optimize traffic light control systems (Barmpounakis, Intelligent transportation systems and
powered two wheelers traffic, 2015) (Wang, 2020) (Litman, 2021). (Li, 2022)proposed an optimal scheme that
reduces the queue delay but the number of passing vehicles remains unchanged.

These methods range from relatively simple cellular automata models (Nagel, 1992) to more sophisticated agent-
based simulations (Chen, 2018)and machine learning approaches (Mileti{\'c}, 2022). These studies have provided
valuable insights into traffic flow (Qadri, 2020)w dynamics and the effects of various traffic light control strategies
on congestion levels.

However, none of this studies tried to investigated the relationship between traffic component through the agent
perspective. Therefore, this study is groundbreaking in the sense of traffic light timing modelling.

65
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

Methodology:-
The model definition is the first part of traffic simulation. It involves giving a mathematical model of the traffic,
describing the global environment (road structures, drivers and their characteristics, traffic light, and rules).

Agent-basedmodeling
A new way of simulating systems with interacting autonomous parts is called agent-based modeling (Cuevas, 2020).
Agents are artificial individuals programmed to perform pre-defined operations (Bemthuis, 2020).

Agent represents the different entities that make up the environment (lattice or multidimensional). Hence they can
either cooperate to form a multi-agent model or work alone. Therefore, the agent's behavior can be described as a
physical system, such as simulations of evacuations, traffic etc.

The main advantage of agents is that they do not need strong mathematical theory or a high level of computing
power. Despite their simple behaviors, they are capable of producing several globally complicated models
(comportments) as a result of the modeling characteristics generated by the interactions of a group of simple agents.
The term "global behavioral models" refers to coherent microscale patterns, such as recognized patterns of temporal,
spatial, and behavioral structure, patterns of distribution, or patterns that are temporally, spatially, and behaviorally
coherent.

Simulation framework
The model defined later corresponds to a traffic situation where vehicle follows each other near a traffic light.
Therefore, the environment is composed by:

Scenarios:
The study focuses on a traffic flow near a junction with a traffic light. The traffic timing is static and we want to
check the impact of traffic light timing, vehicle density, and lane number on the flow. The simulation is conducted
on a road of 20 meters.

Figure 1:- Road section with Traffic light.

66
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

Agent:
To achieve those objectives, we define our agents.

Car agent :
In a simulation context drivers and vehicles could be seen as one entity. However, they are not in real life. Because,
the vehicle simulation is straightforward with speed, and acceleration, the driver is a decision-taker through an
analysis of the traffic situation. Therefore, we assumed that each agent moved toward a certain direction, could
change their behaviors (decisions), and had a set of path solutions. they can be defined as adaptive, reactive,
autonomous, and mobile agents. Each agent is considered as he is going at his own speed and thedecision of
changing lanes is taken by the observation of the next following cells. Although the driver has preferred cells, he
will go to the empty one.

Pedestrian agent:
Pedestrian agents are individuals person who walk through the transportation environment (road). He has to go from
one place to another with a velocity from standing to a maximum speed corresponding to running. The pedestrians
can either be walking, waiting, or halted due to an obstacle. In the context of this paper, the only interactions
considered are the presence of a pedestrian on the road which can yield to an accident and disrupt the traffic.

Traffic Light:
A traffic Light is a road signal device that controls vehicle traffic at an intersection by alternating displaying traffic
state. The traffic light interacts with the car agent in the flowing way red(mandatory stop), and green (vehicle move
on). The transition between the two states is guided by the traffic density. If the traffic density decreases to a given
threshold the traffic becomes red otherwise it stays green. In this paper, the traffic light is supposed to have a
random density from an interval
Simulation:Simulating a single agent on a road section is quite simple since it just consists of acceleration and
deceleration. However, when other agents get involved, it becomes necessary to take into consideration how the
driver evolved and the decision he takes considering other agents. The simulation is done with discrete values,
whereby each iteration is a time step. The initialization value is taken for the Ministry of Transportation. Then the
traffic evolution is investigated through the analysis of travel time, and traffic throughput. Each agent is simulated
by a Class function. The validation is done by real values taken from conductors on busy days. The implementation
of an efficient simulation lies in the definition of a set of rules.

Rules:
The scenario is guided by different rules chosen among the traffic parameters. The traffic light timing is static
therefore the traffic cannot absorb the flow in different situations.

Mathematical model
Since the traffic flow is an interaction between drivers and their environment. In this paper, we combined a set of
mathematical equations describing the actions of considered components.

Considering the aforementioned characteristics of the agents, we will define some guideline rules.

Car followingrules
The car following model used is an extension of Nagel-Schreckenberg (Nagel, 1992) in (Titarmare, 2020) to express
the velocity in different situations. The car following rules depends on the other drivers' behavior. Therefore, there is
cooperation between a car agent and others in the following situations. In each situation, the car adopts a new
velocity Vnew .

Acceleration:
The car accelerates if the space headway is sufficiently large and the preceding vehicle maintains a velocity higher
than the considered vehicle, the car velocity is then between the current one and the maximal velocity given by the
preceding. The lane changing and acceleration will be considered in the next section:

67
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

Braking:
The car brakes if the preceding is lower than the one of the considered vehicle. However, the driver in this situation
maintains a safe space headway sufficient to react to an obstacle. The new velocity is therefore between the current
vehicle and minimum velocity Vmin .

Driving:
When there is no other vehicle, the car chooses its velocity between an interval from a minimum velocity to a
maximal velocity given by the authority

Random velocity:
A random velocity can occur when there is an unexpected event on the road. The only unexpected events that can be
considered here are an accident or when a pedestrian comes too close to the first lane. Since those situations are
discreet events. It's introduced with a binomial negation distribution

Where p is the probability that an accident occurs and q=1-p, k is the number of iteration and n is the number of
success

Lane changing grules

The lane-changing rules are taken from the Nagel-Schrenberg model and state guidelines for vehicular agents when
deciding to move from their current lane to an adjacent lane. The Lane changing occurs in two situations
1. Discretionary Lane Changing: It occurs when the driver sees a queue far away and change lane to go faster
2. Mandatory Lane Changing: This is when the driver has to change lanes because of an obstacle or to follow a
given direction. Considering the speed loss caused by the mandatory lane-changing behavior
3. Before initiating a lane change, a car agent should consider several options:
4. Necessity: A lane change might be triggered by a turn ahead, a slow-moving vehicle in the present lane, or a
planned maneuver to reduce the agent's travel time.
5. Safety: A lane change should be performed only when it is safe to do so. Safety checks should verify that there
is enough space in the target lane and that the speeds of cars in both the current and target lanes are considered.
𝑉𝑛𝑒𝑤 = 𝑚𝑎𝑥⁡ (𝑉𝑛𝑒𝑤 −1 , 0)

Expérimental Design
As shown in figure 1, the experiment is conducted on a road section not far from a traffic light. The road has 3 lanes
and the priority is to keep right unless overtake. The structure of this intersection is such that in addition to the 4
lanes, there is a mini roundabout. Therefore, the static timing is not optimal during the whole day. Throughout the
simulation, the road network acts as a stage for the interactions between vehicles, traffic lights, and the surrounding
environment. It provides the spatial context within which the multi-lane traffic flow and traffic light control
strategies are evaluated

The cells grid is defined by a basic numpy matrix. The number of cars is generated using the initial density and the
number of lanes. the initial car positions are given by the random function of numpy and a loop over the
enumeration of car position fill the grid.

68
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

Figure 2:- Driver decision process.

Basically, the most important part of the evacuation is in the search module. When the traffic light goes from red to
green, the drivers initiate a movement from one cell to the other. However, before he could move, he has to check if
the desired next cell is empty.

Model Validation
The model validation is conducted by checking an obvious assumption, the correlation between traffic density and
road congestion.

Figure 3:- Simulation during 15 seconde.

The simulations show that the traffic flow is smooth when vehicle density is low. In the left Figure there are fewer
empty cells than in right. So the model could be used to improve traffic scenario by adding more realistic

69
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

assumptions (lane changing, pedestrians, and buses) to check the impact of density, light timing, and lane number on
the flow.

Results and Discussion:-

Impact of lanenumber on traffic Congestion

Figure 4:- Impact of lanenumbers on the Traffic.

Figure 4 result on the variation of lane number with a constant density equal to 0.8. The plot show that the average
density function has an inverse variation to the lane number. This is because more lane allows a huge number of
vehicle to cross the junction. Since the result can help traffic regulator to increase the mobility experience, the model
should take consideration of others perspectives like speed limit, number of interaction and light timing. Moreover,
the result is consistent with Texas Transportation Institute who found that increasing the number of lanes from two
to four can reduce traffic congestion by up to 20%.

Impact of Traffic Light Timing on Traffic congestion

Figure 5:- Light timing varies from 30 to 91 s.

70
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

Figure 6:- Light timing varies from 40 to 70 s.

Figure 5 and Figure 6 show the relationship between a green light duration and traffic congestion. It can be seen that
increasing the light timing reduces the traffic congestion. However, the rate improvement decrease over time as
traffic reaches a more balanced state.

Impact of Pedestrian On the Traffic flow

Figure 7:- Impact of Pedestrian on The Flow.

71
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

In Figure 7, we considered a side walker randomly placed. However, this requires drivers to slow down when
approaching them. This results to an increasing probability of accident at each iteration. Accidents disrupt traffic
flow and demonstrate the impact of unexpected events on congestion. By incorporating these assumptions, we are
more comprehensive about the reaction of traffic on unexpected events.

Travel Time Comparison

Figure 8:- Travel time comparison with distributions.

Figure 8 is a scenario conducted when the traffic density varies from 0,1 to 1,0. It can be shown that the static traffic
light reacts better when the traffic density is low. This because the drivers have more time to adjust theirs speed to
the traffic light. And the dynamic traffic tends to behaves like a static traffic light.

Figure 9:- Travel distribution.

72
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

In Figure 9, we have a heavy jam, where we can see that the dynamic traffic light travel time is less than the one of
static traffic light. This is because the driver agent are better used by the traffic light agent to regulate the mobility.

Conclusion:-
This paper is a multi-agent-based model where agent rules are based on the cellular automata equation. The
simulation shows that this pattern is a good representation of traffic congestion. However, the model could be
improved with streaming data provided by the sensor whereby the density of the different agents would be more
accurate. Furthermore, the slightly difference between dynamic and static travel time can be due to the use of
empirical data during simulation.

However, this result may prompt road planners to consider implementing other technologies such as fly-over or fully
connected traffic light to reroute the traffic in case of emergency.

Summary of results:
Our study demonstrated that the use of Agent Based Model gives a better understanding of the microscopic
interactions between different agents. The different scenarios showed that dynamic traffic light deals better with
traffic jam

Practical implications:
Thepractical implications of our research are numerous. Firstly, our approach can contribute to more enhance urban
mobility, reducing the time spent at traffic light. It can also help traffic planner to size up different path when there
is in emergency.

Future prospects:
For future research, there are several interesting avenues to explore. Firstly, implementing the solution in real
situation. Then, the solution could be enhancing with a Deep Learning prediction to better predict the different
timing.

References Bibliographies:-
[1]Barmpounakis, E. N. (2015). Intelligent transportation systems and powered two wheelers traffic. IEEE
Transactions on Intelligent Transportation Systems, 908--916.
[2] Barmpounakis, E. N. (2016). Unmanned Aerial Aircraft Systems for transportation engineering: Current practice
and future challenges. International Journal of Transportation Science and Technology, 111--122.
[3] Bemthuis, R. a.-E. (2020). Using agent-based simulation for emergent behavior detection in cyber-physical
systems. IEEE.
[4] Berhanu, Y. a. (2023). Examining Car Accident Prediction Techniques and Road Traffic Congestion: A
Comparative Analysis of Road Safety and Prevention of World Challenges in Low-Income and High-Income
Countries. Journal of Advanced Transportation.
[5] Chen, Y. a. (2018). An agent-based stochastic Occupancy Simulator. Building Simulation: Springer, 37--49.
[6] Cuevas, E. (2020). An agent-based model to evaluate the COVID-19 transmission risks in facilities. Computers
in biology and medicine.
[7] Huang, H. a. (2014). AffectRoute--considering people’s affective responses to environments for enhancing
route-planning services. Taylor \& Francis, 2456--2473.
[8] Huang, Z. a. (2023). Urban traffic congestion in twelve large metropolitan cities: A thematic analysis of local
news contents. International Journal of Sustainable Transportation, 592--614.
[9] Kawajiri, K. a. (2020). Lightweight materials equal lightweight greenhouse gas emissions?: A historical analysis
of greenhouse gases of vehicle material substitution. Journal of Cleaner Production.
[10] Li, W. a. (2022). Human-vehicle intersection traffic lights timing optimization research. Journal of advanced
transportation.
[11] Litman, T. (2021). Well Measured-Developing Indicators for Sustainable and Livable Transport Planning-5
March 2021. VTPI: Victoria Transport Policy Institute.
[12] Mileti{\'c}, M. a. (2022). A review of reinforcement learning applications in adaptive traffic signal control. IET
Intelligent Transport Systems, 1269--1285.
[13] Nagel, K. a. (1992). A cellular automaton model for freeway traffic. Journal de physique I, 2221--2229.

73
ISSN: 2320-5407 Int. J. Adv. Res. 11(11), 64-74

[14] Nedim, K. a. (2023). Advanced Traffic Management System in the Function of Improving Mobility in the City
of Sarajevo. Springer.
[15] Nigam, N. a. (2023). A Review of Different Components of the Intelligent Traffic Management System
(ITMS). Symmetry: MDPI, 583.
[16] Titarmare, A. S. (2020). Analysis of traffic flow at intersection to avoid accidents using Nagel-Schreckenlerg
(NS) model. IEEE.
[17] Wang, H. a. (2020). Optimizing signal timing control for large urban traffic networks using an adaptive linear
quadratic regulator control strategy. IEEE Transactions on Intelligent Transportation Systems, 333--343.
[18] Zhu, F. a.-Y. (2019). Parallel transportation systems: Toward IoT-enabled smart urban traffic control and
management. IEEE Transactions on Intelligent Transportation Systems, 4063--4071.

74