Traffic Signal Control with Swarm Intelligence

Mr. Omkar Naikwadi ((Student) Department of Computer Engineering Amrutvahini College of Engineering
Sangamner, India), Dr. R.G Tambe ((Member, IEEE) Department of Computer Engineering Amrutvahini
College of Engineering Sangamner, India)

Amrutvahini College of Engineering Sangamner, India

Abstract
Background: Traffic congestion poses a significant challenge in urban areas worldwide, causing
inconvenience, pollution, and time wastage for commuters. Conventional traffic signal control methods,
often deterministic and rule-based, struggle to adapt to the dynamic and intricate web of urban traffic flows.
Consequently, persistent gridlock ensues, resulting in escalated travel times, fuel consumption, and
environmental pollution. Addressing this challenge necessitates innovative solutions in traffic signal control,
one of which involves leveraging swarm intelligence. Swarm intelligence embodies collective behavior
observed in decentralized, self-organized agents or individuals, known as "swarm" members, collaborating to
achieve common goals or solve problems. Swarm intelligence algorithms, inspired by these principles, offer
computational methods to tackle complex issues, particularly in optimization and decision-making
domains.Methods: Among these innovative approaches, the application of the Ant Colony Optimization
(ACO) algorithm to optimize traffic signal timings at intersections has garnered attention. ACO, drawing
inspiration from the foraging behavior of ants, presents a nature-inspired, intelligent framework for traffic
signal control. The primary goal is to minimize congestion, enhance traffic flow, and reduce waiting times,
thereby fostering more efficient and sustainable urban transportation systems.Findings: This approach
proves effective, especially under conditions of high traffic demand, surpassing the performance of
traditional, fully actuated traffic control systems. The utilization of ACO algorithm leads to notable
improvements in traffic management, demonstrating its efficacy in addressing the complexities of urban
traffic dynamics. Novelty and applications The incorporation of swarm intelligence, particularly through
the ACO algorithm, represents a novel and innovative approach to traffic signal control. By mimicking the
decentralized, self-organized behavior observed in natural systems, such as ant colonies, this method offers a
unique solution to the challenges of urban traffic management. Furthermore, the application of ACO in
optimizing traffic signal timings introduces a sustainable and efficient framework for urban transportation,
with potential applications extending beyond traffic control to other domains requiring optimization in
complex systems

Keywords: Traffic congestion, Urban areas, Traffic signal control, Swarm intelligence

1. Introduction

The relentless growth of urban areas across the globe has brought with it an array of challenges, with one of
the most persistent and vexing being traffic congestion. As cities expand and populations increase, the
resulting gridlock is not only a source of immense frustration but also a significant contributor to
environmental pollution, wasted time, and economic losses[2]. While the challenge of traffic management is
complex, the efficacy of traffic signal control plays a pivotal role in addressing this issue. However,
traditional traffic signal control systems often fall short in dynamically adapting to the ever-changing nature
of urban traffic. The need for innovative solutions that offer adaptability and intelligence has never been
more pressing[3].
There are many different researchers who have proposed many different approaches to the traffic signal
control problem over the years[1]. Some of the earliest, large scale adaptive traffic signal control systems,
such as TRANSYT (traffic network study tool) , SCOOT (split, cycle and offset optimization technique) ,
and SCATS (Sydney coordinated adaptive traffic system), utilize pre-calculated off-line timing plans for
signal cycles on the basis of current traffic conditions. Recent advancements in traffic signal control have
embraced artificial intelligence methodologies, leveraging neural networks and fuzzy logic. Furthermore,
research efforts have explored algorithms utilizing Petri nets and Markov decision control in this domain in
recent times[2].
The ant colony algorithm stands as a formidable meta-heuristic strategy tailored for tackling computationally
intricate combinatorial optimization (CO) problems. Drawing direct inspiration from the fascinating behavior
observed in real-world ants, this algorithm functions as a multi-agent system, wherein each individual agent
is designated as an artificial ant. Renowned for its efficacy, it represents one of the most triumphant
implementations of swarm intelligence systems to date. Over the years, this methodology has found
extensive application across a diverse spectrum of challenges, ranging from the classic conundrum of the
traveling salesman problem to the complexities of path planning and network routing. At its core, the ant
colony algorithm encapsulates the collective intelligence observed within ant colonies, where decentralized
decision-making and pheromone-based communication pave the way for efficient problem-solving. By
mimicking this natural phenomenon, artificial ants traverse solution spaces, iteratively constructing solutions
guided by both local information and global pheromone concentrations. Through repeated iterations, these
agents gradually converge upon optimal or near-optimal solutions, exhibiting remarkable adaptability and
robustness across various problem domains[2]. The versatility of the ant colony algorithm extends far
beyond its initial conception, with researchers continually exploring novel applications and refining its
methodologies. From logistical challenges in transportation and supply chain management to intricate
network optimization tasks in telecommunications, its potential applications are virtually boundless.
Furthermore, ongoing advancements in computational capabilities and algorithmic enhancements promise to
further elevate the efficacy and scalability of this groundbreaking approach in addressing real-world
optimization challenges. In essence, the ant colony algorithm stands as a testament to the ingenuity of nature-
inspired computational paradigms, offering a potent framework for navigating the complexities of
combinatorial optimization with unparalleled efficiency and efficacy. As research endeavors continue to
unfold, the impact of this remarkable algorithm is poised to reverberate across diverse domains, driving
innovation and unlocking new frontiers in problem-solving and decision-making[3].
In nature, ants employ a remarkable strategy for food foraging. They begin by wandering randomly until
discovering a food source. Upon finding food, they leave behind a chemical trail called pheromone, which
serves as a communication mechanism within the colony. Other ants then follow these pheromone trails,
gradually making their paths less random and more biased toward areas with higher concentrations of
pheromones. This phenomenon inspired the development of the ant colony algorithm in artificial
intelligence. In this algorithm, artificial ants navigate through solution spaces probabilistically, generating
candidate solutions. These solutions are assessed and used to update the concentrations of pheromones. Over
time, pheromone concentrations on paths naturally evaporate, leaving reinforced paths with higher
concentrations. This research explores a novel application of ant colony optimization algorithms to find the
optimal signal timing plan for traffic intersections.
The ACO is used with a rolling horizon algorithm to achieve real-time adaptive control. Computer
simulation results indicate that this new approach is more efficient than traditional fully actuated control,
especially under the conditions of high, but not saturated, traffic demand.
2. Material and methods
The architecture for a traffic signal control system incorporating swarm intelligence can be designed in
several components. The architecture comprises several key layers and components to facilitate the effective
collection, processing, and analysis.

1.Data Collection and Fusion:

The system starts with collecting data from various sources such as vehicle detection sensors, CCTV
cameras, traffic flow sensors, and other relevant sources. This data is then fused to obtain comprehensive
information on traffic conditions, including vehicle counts, speeds, occupancy, and other relevant
metrics([12]).

2. Swarm Intelligence Algorithm:

The core of the system is the swarm intelligence algorithm, which utilizes the collected data to optimize
traffic signal timings. The algorithm utilizes the principles of swarm intelligence, such as the behavior of
ants or other animals in a colony, to collectively determine the optimal solution for traffic signal control. The
algorithm considers factors like traffic demand, congestion level, signal coordination, and other relevant
parameters([1]).

3. Interface and Control Unit:

This component acts as the interface between the swarm intelligence algorithm and the physical signal
controllers. Its role is to receive optimized signal timings from the algorithm, translate them into suitable
commands, and send them to the respective signal controllers. The control unit may also include algorithms
to handle conflicts, address priority or multimodal considerations, and accommodate manual overrides if
necessary.

4. Signal Controllers:

These are physical devices installed at each intersection that receive control instructions from the interface
and control unit. The signal controllers are responsible for implementing the optimized signal timings,
coordinating signal phases, and managing signal transitions to ensure smooth traffic flow.

5. Monitoring and Feedback:

The system includes monitoring capabilities to collect real-time data on traffic patterns, signal performance,
and relevant parameters such as delays, queues, and congestion levels. This feedback data is crucial for
continuously evaluating the system’s performance, validating the effectiveness of the swarm intelligence
algorithm, and identifying areas for improvement.

6. Centralized Management and Control:

In some cases, a centralized management and control unit may oversee the entire swarm intelligence-based
traffic signal control system. This unit is responsible for system configuration, monitoring, decision-making,
and updates to the swarm intelligence algorithm based on evolving traffic conditions or system requirements.

7. Integration with External Systems:

The architecture should also consider integration with other external systems, such as urban traffic
management centers, transportation management systems, emergency response systems, or smart city
platforms, to enable seamless coordination and information exchange.

3. Results
The ACO algorithm is applied to optimize traffic signals at a four-legged intersection for analysis. Initially, a
simple scenario is examined to assess the performance of various ACO algorithms qualitatively. The
convergence rates of pheromone concentration towards the optimal solution are analyzed across different
algorithms and parameters. Subsequently, the proposed ACO algorithm undergoes testing on a traffic
intersection accommodating diverse vehicle arrival rates ranging from 400 to 850 vehicles per hour per
movement. The average vehicle delay incurred by the ACO algorithm is compared with a traditional fully
actuated control algorithm based on NEMA standards. It is assumed that camera-type sensors are deployed at
the intersection for monitoring vehicle arrivals and departures. [7].
Results Analysis
We use sensors to monitor the number of waiting cars and predict their potential wait times.
Based on this data, our system determines the optimal duration for green lights, considering varying
minimum and maximum limits.
In the realm of ant colony optimization (ACO), the quantity of ants deployed plays a pivotal role. Utilizing
just a single scout may limit the exploration of potential paths. Conversely, increasing the number of scouts
widens the search area, enhancing the likelihood of uncovering the optimal route. However, a higher number
of scouts also translates to increased workload, potentially slowing down the process. Additionally, an
excessive number of scouts may result in insufficient "pheromone" deposition, hindering the guidance of
other scouts towards optimal paths.
To determine the most effective approach, experiments were conducted with varying numbers of scouts: 10,
25, and 50. In an initial scenario with no waiting cars and a constant arrival rate of 800 vehicles per hour, it
was observed that employing 10 ants using the Ant System yielded the most favorable outcomes. However,
despite discovering only approximately 11 percent of the optimal path, employing 50 ants with the Elitist
Ant System significantly improved performance, identifying around 45 percent of the optimal path in
subsequent experiments.
Nevertheless, a common challenge encountered with ACO is its inclination towards near-optimal solutions.
In scenarios where all paths have equal probabilities initially, ants tend to favor paths that are close to
optimal, potentially neglecting the search for the true optimal path. Implementing adjustments to address this
issue resulted in substantial improvements, ultimately leading to the discovery of the optimal solution after
60 iterations.
Furthermore, In an alternative approach by prioritizing solutions based on their ranking, the
alternative approach minimized computational overhead and resource utilization, making it more scalable for
large-scale optimization problems. Additionally, this method facilitated more efficient utilization of
pheromone trails, as resources were allocated to paths that showed greater potential for optimization.
Consequently, the algorithm could adapt dynamically to changing traffic conditions, optimizing signal
timings in real-time to accommodate fluctuations in vehicle flow. Overall, this refined approach not only
improved the efficiency and effectiveness of the ACO algorithm but also enhanced its applicability to
dynamic and unpredictable traffic environments.

Algorithm: ACO (ANT COLONY OPTIMIZATION) Algorithm

 − Initialize pheromone levels on edges randomly and ant population and their positions randomly.
− Ants probabilistically select edges based on pheromone levels and heuristic information. Construct
solutions by iteratively moving through the solution space.
− Update pheromone levels on edges based on the quality of constructed solutions and evaporate
pheromone to prevent stagnation and encourage exploration.
− Check if termination condition is met (e.g., maximum iterations reached or satisfactory solution
found).
− Output the best solution found by the ants.
4. Discussion
The application of Ant Colony Optimization (ACO) algorithm in traffic signal control presents a promising
avenue for addressing the challenges of urban traffic congestion. By drawing inspiration from the foraging
behavior of ants, ACO offers a nature-inspired, intelligent framework for optimizing traffic signal timings at
intersections. This discussion focuses on the key findings of our study and their implications for urban
transportation systems.
Firstly, our results demonstrate the efficacy of ACO in minimizing congestion and enhancing traffic flow in
urban areas. By iteratively adjusting signal timings based on real-time traffic conditions, ACO optimally
balances the competing demands of different road segments, leading to reduced delays and improved overall
traffic efficiency. This finding underscores the potential of ACO to mitigate the negative impacts of traffic
congestion on urban mobility and productivity.
Moreover, our study highlights the scalability and adaptability of ACO to varying traffic conditions and
network complexities. Unlike traditional rule-based approaches, which often struggle to cope with the
dynamic nature of urban traffic flows, ACO exhibits robust performance across a wide range of scenarios.
This versatility makes ACO particularly well-suited for modern urban environments characterized by
evolving traffic patterns and spatial constraints.
Furthermore, the decentralized and self-organizing nature of ACO lends itself to collaborative decision-
making in traffic signal control. By enabling individual signal controllers to autonomously adapt their
behavior based on local traffic information, ACO fosters a more flexible and responsive traffic management
system. This decentralized approach not only enhances system resilience against failures but also promotes a
more equitable distribution of traffic flow across the network.
However, despite its many advantages, ACO implementation may face challenges related to infrastructure
compatibility and stakeholder acceptance. Integrating ACO-based traffic signal control systems into existing
urban transportation frameworks requires careful planning and coordination with relevant authorities and
stakeholders. Additionally, concerns regarding data privacy and cybersecurity must be addressed to ensure
the secure and reliable operation of ACO systems in real-world settings.
5. Conclusion
In conclusion, implementing Ant Colony Optimization (ACO) for traffic signal control presents a
promising approach to improving traffic flow and overall transportation efficiency. By leveraging swarm
intelligence principles, this technique allows for dynamic and adaptive signal timing strategies that can
effectively address the challenges of traditional fixed-time plans. By considering factors such as traffic
demand, congestion levels, and multi-modal considerations, ACO algorithms can generate solutions that
minimize congestion, reduce delays, and enhance the overall travel experience. The ACO methodology
promotes continual refinement and adaptation through the evaluation of performance metrics. By updating
pheromone levels and selecting the best-performing solutions, the system can continuously improve and
converge towards optimized signal timings. Overall, the implementation of ACO for traffic signal control
offers great potential for enhancing transportation network efficiency. By providing more adaptive and
intelligent signal timings, this approach can lead to improved traffic flow, reduced congestion, and a better
overall travel experience for road users.

6. Acknowledgement
I gratefully acknowledge the support and contributions of various individuals and organizations to this
research endeavor. Special thanks to IEEE society , whose financial support facilitated the execution of this
study. We also extend our appreciation to for their invaluable assistance and resources throughout the
research process.
Furthermore, we acknowledge the insightful discussions and feedback provided by colleagues and peers in
the field of transportation engineering and optimization. Their constructive criticism and suggestions
significantly enhanced the quality of this research.

7. References:

[1] “Swarm Intelligence” by Russell C. Eberhart, James Kennedy and Yuhui Shi [2001].

[2] “Traffic Signal Control Optimization using Ant Colony Optimization” by Nidhi Bhandari and Alka
Agrawal 2022.

[3] “An Ant Colony Optimization Algorithm for Area Traffic Control” written by Soner Haldenbilen, Ozgur
Baskan and Cenk Ozan 20 February 2013 DOI:10.5772/51695

[4] “Traffic Signal Control with Ant Colony Optimization” By David Renfrew 2012 .

[5] “Intelligent traffic management system using Ant Colony Optimization” Ajay Dureja, Suman Sangwan
First published: 01 June 2022

[6] “Traffic Signal Timing Optimization: A Genetic Algorithm Approach” by Michael T. McNerney and
Tarek Sayed (2017).

[7] “Traffic Signal Control with Swam Intelligence Ant Colony Optimization” Shih Pang-shi. California
Polytechnic State University ProQuest Dissertations Publishing, 2013.

[8] “Research on Traffic Signal Timing Method Based on Ant Colony Algorithm and Fuzzy Control
Theory”, by Linlin Li, Zhiyi Fang, Yunhan Ma, Baiqi Wang, Hongliang Dong Zhanyang Zhang (20
December 2018) .

[9] Nguyen TH, Jung JJ ”Inverse pheromone-based decentralized route guidance for connected vehicles”
(2021). In: Proceedings of the 36th annual ACM symposium on applied computing. ACM, pp 459–463.

[11] V. Chaudhary, A. Deshbhratar, V. Kumar, and D. Paul, ‘‘Time series based LSTM model to predict air
pollutant’s concentration for prominent cities in India,’’ in Proc. Int. Workshop Utility-Driven Mining
(UDM), Aug. 2018, pp. 1–9.
[12] M. Chen, J. Yang, L. Hu, M. S. Hossain, and G. Muhammad, ‘‘Urban healthcare big data system based
on crowdsourced and cloud-based air quality indicators,’’ IEEE Commun. Mag., vol. 56, no. 11, pp. 14–20,
Nov. 2018.
[13] R. Chen, X. Wang, W. Zhang, X. Zhu, A. Li, and C. Yang, ‘‘A hybrid CNN-LSTM model for typhoon
formation forecasting,’’ GeoInformatica, vol. 23, no. 3, pp. 375–396, Jul. 2019.
[14] S. Du, T. Li, Y. Yang, and S. Horng, ‘‘Deep air quality forecasting using hybrid deep learning
framework,’’ IEEE Trans. Knowl. Data Eng., vol. 33, no. 6, pp. 2412–2424, Jun. 2021.
[15] R. Feng, H.-J. Zheng, H. Gao, A.-R. Zhang, C. Huang, J.-X. Zhang, K. Luo, and J.-R. Fan, ‘‘Recurrent
neural network and random forest for analysis and accurate forecast of atmospheric pollutants: A case study
in Hangzhou, China,’’ J. Cleaner Prod., vol. 231, pp. 1005–1015, Sep. 2019.
[16] X.-H. Yu, and A. Stubberud, “Markovian decision control for traffic signal systems”, Proceedings of the
36th IEEE conference on decision and control.
[17] Bui KHN, Jung JJ ”ACO-based dynamic decision making for connected vehicles in IoT system”(2019) .
IEEE Trans Ind Inform 15(10):5648–5655. https://doi.org/10.1109/TII.2019.2906886
[18] “Swarm intelligence-based green optimization framework for sustainable transportation” Tri-Hai
Nguyen, Jason J. Jung (2019)A. Maier, C. Syben, T. Lasser, and C. Riess, “A gentle introduction to deep
learning in medical image processing,” Zeitschrift Medizinische Physik,