Advanced Technology for Improved Time

Management in Traffic Signals

Abhinandan Hore Asif Khan

M. Tech, Department of CSE, M. Tech , Associate Professor , Department CSE
Greater Noida Institute of Technology (Engineering Institute), GNIOT Engineering Institute,
Noida, India Greater Noida, India
Email ID – [email protected] Email ID – [email protected]

Abstract
The goal of the time optimization for traffic signal performance presented in this paper is to minimize both
traffic and time in the most convenient manner. A smart convergence framework model is developed in this
paper. The study of the model includes with the techniques of artificial intelligence and machine learning
algorithms approaches. This research is to generally way to implement in a real world where there is a big
problem of traffic congestion in metro cities and towns. This research will leads to a betterment of the people
in terms of transport. The fixed time system of signals is typically used to control automated signals. Traffic
control systems with fixed time systems operate the signals at predetermined intervals, which reduces time
wasted when there are fewer vehicles in a line. In order to improve adaptability and accuracy, this study
combines a number of approaches, including AI and metaheuristics. It combines multiple approaches,
Genetic Algorithms with Fuzzy Logic. It leverages strengths of multiple algorithms. Hybrid methods have
shown that improvements of up to 40% in reducing delays. In my research, I tried to use hybrid models,
which combine fuzzy logic and genetic algorithms, to effectively improve the efficiency of traffic control
signals by optimizing their time. Instead of traditional classification/regression accuracy, these algorithms'
effectiveness or "accuracy" is typically measured in terms of their capacity to optimize traffic metrics like
average delay, queue length, and throughput. Depending on the city, traffic patterns, and sensor fidelity, real-
world accuracy may vary. In this research, we have shown simulation of hybrid model which is a combination
of Genetic algorithm and fuzzy logic for traffic signal control.

Keywords: Machine Learning (ML), Traffic signal optimization, Genetic Algorithm, Multi-Agent Systems,
Fuzzy Logic, Hybrid Genetic Algorithm- Fuzzy.

I. INTRODUCTION

Traffic signal control is one of the essential elements in intelligent transport systems, which manage local networks
from end to end. The objectives of traffic signal control system design are primarily to minimize waiting time of the
vehicles and the best possible traffic flow on the roads. These are greatly dependent on signal-timing plans, which
allocate green time to different traffic movements under safety and efficiency restrictions. Optimization theory has
been employed to design signal-timing plans traditionally in solving the problem. Yet, traffic flow through an
intersection is highly complex, and traffic demand working from day to night is unpredictable and dynamic. Thus,
methods and optimization models must keep these features in mind to model the minute changes in real-time traffic
flow successfully.
Isolated traffic signal flow control used for fixed-time optimization is implemented today in the traffic signals. The
goal is to determine the optimal cyclic plans that have minimal waiting time at the intersection with a fixed cycle
length. Various types of signal controls, such as multi-ring, phase-based, and phase/split-based control with or
without protected left-turns, are also addressed. There are several various types of traffic flow, including real-time
demand and undersaturated/saturated flow, where there is enough or too little green time available for all the
movements. It is proven that the single models are worse than the hybrid Genetic Algorithm-Fuzzy model. because:
 (i) It combines the real-time adaptability of fuzzy logic with the global optimization capabilities of GA.
(ii) It consistently outperforms standalone systems across key traffic performance metrics.
Urban traffic congestion has become a significant problem with the rate of urbanization and increased number
of cars on the roads. Efficient traffic management is required to alleviate delays, conserve fuel, and enhance the
travel experience. One of the most significant parts of traffic management is enhancing the flow of vehicles on
the roads by enhancing the performance of the traffic system. Getting traffic to move more smoothly is the main
objective of optimizing traffic light timing. This means minimizing delays, preventing traffic lines from
becoming too long, and getting more cars through intersections. Conventional traffic light systems, which
operate on a fixed schedule, usually can't cope with the constantly changing conditions of traffic very well.
This can lead to traffic congestion and wasted time. That is why more advanced techniques, such as machine
learning, traffic simulation, and adaptive control systems, are becoming increasingly popular for combating
these problems.
This research paper aims to find ways of making traffic lights function more efficiently. It employs computer
tools and data-dependent methods of making the way traffic systems function more efficient. The study
considers crucial aspects such as variations in traffic, rush hour traffic congestion, and pedestrian needs.
Through this, the research aims to assist in designing smart transportation systems that are efficient and
environmentally friendly.
The aim of this research paper is to discover new means of enhancing the functioning of traffic signals. It analyzes
adaptive control systems, real-time information, and advanced computer tools. The research investigates advanced
techniques such as traffic simulation models, AI-based algorithms, and multi-objective optimization methods. It
attempts to overcome the limitations of conventional systems. It also ultimately seeks to contribute towards the
development of intelligent, efficient, and environmentally friendly traffic signal systems that align with the concept of
smart, greener cities. The study confirms that traffic signal timing optimization is not only a technical solution but
also an important step in the direction of safer, more efficient, and eco-friendly city transportation systems.

II. TIME OPTIMIZATION ALGORITHMS

It has become of utmost importance to enhance the function of traffic lights to save time. We should find means of
accelerating and facilitating transportation as towns become increasingly more crowded. In this review, we examine
the primary research and approaches in doing so, organizing them into three categories: standard approaches,
simulation-based models, and state-of-the-art machine learning methods. In essence, in this research paper we have
examined two algorithms in terms of their efficiency and effectiveness.
Genetic Algorithm (GA) for Traffic Signal Control Time Optimization
Genetic Algorithms (GAs) are optimization algorithms modeled with the process of natural selection. They are mainly
effective in problem solving, non-linear, complex and multi-objective problems like traffic signal timing optimization. It
is generally used for solving complex optimization problems like the duration of traffic signal, where the goal is to
minimize congestion, reduce travel time, or optimize fuel consumption.
Working in Traffic Signal Optimization :
1. Encoding the Problem: Each representation of a chromosome is a potential solution, like signal timing plans for every
stage of a traffic light.
(i) The durations of the traffic signal phases, i.e., the green, yellow, and red periods, are represented as a
chromosome a sequence of values.
(ii) Example chromosome: [40, 10, 50, 35] (green = 40s, yellow = 10s, red = 50s, cycle length = 110s).
2. Initialization: Create a starting population of chromosomes (potential signal timing plans) at random or heuristically.
Solutions create a population of chromosomes at random.
3. Fitness Function: Assess each chromosome's efficiency in terms of performance factors like, Measures how good a
solution is based on factors such as:
(i) The average amount of time spent waiting for a vehicle.
(ii) Queue length.
(iii) Traffic flow throughput.
(iv) Minimization of delay.
(v) Minimizing average waiting time.
(vi) Reducing queue lengths.
(vii) Maximizing traffic flow.
Performance Metrics: There are no standard "accuracy" scores for Genetic Algorithms but rather are measured in
terms of traffic performance metrics:
(i) Reduction of Average Waiting Time: 20-30% compared to fixed-timing systems.
(ii) Traffic Flow Improvement: 15-30%.
(iii) Fuel Consumption Reduction: Up to 30%. Real-World/Simulation Results:

Fitness = 1
𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝐷𝑒𝑙𝑎𝑦

A. Fuzzy Logic for Traffic Signal Control Time Optimization

Fuzzy logic is a useful method to deal with uncertainties in traffic light control, in which the inputs and outputs are
not crisp but on a scale of degrees of truth (e.g., "low," "medium," "high"). This system utilizes a set of rules and
membership functions to calculate the ideal length of green, yellow, and red lights from real-time traffic
information. Fuzzy logic is a rule-based system capable of dealing with uncertainties and imprecise information.
Since real-world traffic data often include uncertainties (e.g., uncertainty in vehicle arrivals and weather conditions),
it is especially suitable for traffic signal control.
Working of fuzzy logic standalone model in traffic control
1. Input Variables:
(i) Traffic flow rate (e.g., cars/min).
(ii) Queue length (e.g., number of vehicles on a queue).
(iii) Waiting time of vehicles at signals.
2. Fuzzification:
(i) Transforms crisp input data into fuzzy sets based on linguistic variables such as:
(ii) Low, Medium, High (for traffic flow rate).
(iii) Short, Moderate, Long (depending upon the queue length).
(iv) Map crisp inputs (e.g., the number of automobiles, time in seconds) to fuzzy sets through membership
functions.

3. Rule Base:
(i) A collection of IF-THEN rules formulated by experts. For instance:
(ii) IF traffic flow is High AND queue length is Long THEN green signal duration is Long.
(iii) IF traffic flow is Low AND queue length is Short THEN green signal duration is Short.
4. Inference Engine:
(i) Analyzes the rules and decides to what extent each rule holds.
(ii) Combine input fuzzy sets and apply rules to produce output fuzzy sets.
5. Defuzzification: Translates the fuzzy output back to a crisp value (e.g., 30 seconds of green signal).
6. Output Variables: Duration of the green light and signal priority adaptation.
Performance Metrics: The performance tends to be analyzed with the aid of simulation tools or actual experimentation.
Measures incorporate shorter waiting time, enhanced flow of traffic, and rate of vehicle flow .The following are the general
outcomes: Accuracy in mimicking real-world scenarios: (depending on the quality of the rule base and input data).
(i) Reduction of average waiting time: 15-22%. Improvement of flow of traffic: 20-22%.
(ii) Travel time savings: 15-22%.
(iii) Improvement in throughput: 15-22%.
(iv) Reduction in fuel consumption and emissions: Up to 20% in some studies.

III. METHODOLOGY
Hybrid Model of Genetic Algorithm (GA) and Fuzzy Logic for Traffic Signal Control
Genetic Algorithm and Fuzzy Logic, when combined, create a highly effective hybrid approach that leverages the
strength of each in optimizing traffic signal timings. Consider Genetic Algorithm as the overall problem solver that
zeros in on the optimum fuzzy rules or membership functions. Meanwhile, Fuzzy Logic handles the moment-to-
moment decisions necessary for traffic signal control. These pairing ensures both adaptability and optimal
performance. Simply put, combining Genetic Algorithm with Fuzzy Logic allows us to utilize the strengths of both
worlds. An adaptive traffic signal control system that is both highly effective outperforms what each method could
do in isolation by fine-tuning fuzzy logic elements, like modifying membership functions or rule sets.
Working of the Hybrid Genetic Algorithm-Fuzzy Approach
1. Fuzzy Logic System It regulates and governs traffic system control through:
(i) Input Factors: Traffic stream, line length, waiting time.
(ii) Output Factors: Signal timings (e.g., green length).
Rule based: It consists of pre-defined IF-THEN rules. Rule illustration:
(i) IF traffic flow on the roads is High AND queue length duration is Long,
(ii) THEN green signal time duration is Long.
2. Fuzzy Logic Framework: Define fuzzy input variables, such as:
(i) Traffic volume (e.g., Low, Medium, High).
(ii) Queue length (e.g., Short, Moderate, Long).
(iii) Waiting time (e.g., Short, Medium, Long).
Define fuzzy output variables: Signal timings (e.g., Green Duration: Short, Medium, Long).
3. Fitness Function: The performance of a fuzzy logic setup is generally estimated by the fitness function on the
following parameters:
(i) An average reduction in delay.
(ii) Traffic throughput improvement.
(iii) Reducing queue lengths.
(iv) Assess performance based on traffic measures:
(v) Constant average delay, queue length, or throughput. Example fitness function:

1
𝐹𝑖𝑡𝑛𝑒𝑠𝑠 = 𝑤1 ∗ + 𝑤2
𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑊𝑎𝑖𝑡𝑖𝑛𝑔 𝑇𝑖𝑚𝑒
∗ 𝑇ℎ𝑟𝑜𝑢𝑔ℎ𝑝𝑢𝑡 − 𝑤3
∗ 𝑄𝑢𝑒𝑢𝑒 𝐿𝑒𝑛𝑔𝑡ℎ

Accuracy and Effectiveness of Genetic Algorithm- Fuzzy Hybrid - Performance Metrics:

(i) Reduction in Average Waiting Time: 30- 50%.
(ii) Improvement in Throughput: 30-40%.
(iii) Reduction in Queue Length: Up to 40%.
(iv) Fuel Consumption Reduction: Up to 35%. Use Case Example:
1. Inputs:
(i) Traffic flow = 120 vehicles/hour.
(ii) Queue length = 15 vehicles.
2. Fuzzy Output: Set green light time to 40 seconds. GA Optimization: Optimizes fuzzy rules
to favor essential traffic lanes, cuts waiting time by 40%.
 Reduction Improvement in
System
in Delay Throughput
Fixed -
Time Control
Baseline Baseline
Fuzzy
Logic Only 15-20% 15-18%
Genetic
Algorithm Only
20-25% 20-22%
Genetic
Algorithm
- Fuzzy
Hybrid 30-50% 30-40%

Table I. Systems using different framework in terms of reduction in delay and improvement throughput

IV. RESULTS AND DISCUSSION

1. Standalone Fuzzy Logic:
(i) Decreases waiting time and queue length by 20-30% over fixed-time signals.
(ii) Limited performance due to static fuzzy rules and membership functions.
2. Standalone GA:
(i) 10% greater throughput than genetic algorithm alone.
(ii) Merges the adaptability of fuzzy logic with the global optimization of GA.
Accuracy Comparison Metric: Reduction in Average Waiting Time
(i) Fuzzy Logic: 15-25% improvement.
(ii) GA: 20-30% improvement.
(iii) GA-Fuzzy: improvement of 40-50%.
Metric: Throughput Improvement
(i) Fuzzy Logic: 10-20% improvement.
(ii) GA: 20-30% improvement.
(iii) GA-Fuzzy: 30-40% improvement.
(iv) logic, with up to 40% reduction in waiting time.
(v) Lacks adaptability to real-time changes in traffic.
Simulations have shown 30-50% improvements in traffic delay compared to fixed-time control and 10- 20% better
performance than standalone fuzzy or Genetic Algorithm systems. Simulations of hybrid systems have shown: Up to
45% reduction in delays at peak hours. Better adaptability to dynamic traffic conditions compared to standalone
fuzzy or Genetic Algorithm systems.

V. CONCLUSION
A combination approach that utilizes the strengths of both Fuzzy Logic and Genetic Algorithm (GA)
is employed to maximize traffic signal timings. GA is applied for global optimization, identifying the optimal fuzzy rule
sets or membership functions, while fuzzy logic manages real-time decision-making for traffic signal control.
The integration ensures flexibility, adaptability and efficiency for the optimized performance.
The combination of Genetic Algorithm (GA) and Fuzzy Logic provides the efficiency and effective of better and
powerful techniques When we bring these two approaches together, we end up with a traffic signal control system that's
not just smart, but also incredibly responsive, surpassing what either method could accomplish solo. This hybrid
approach enables betterment to the people in the modern world while bringing live to the traffic control management.
Optimizes signal timings better than fuzzy logic, leading to a potential 40% decrease in waiting time. Surpasses the
performance of individual models, reducing the waiting time by 50% and enhancing throughput by 10% when compared
to using only genetic algorithm.
 Average
Queue
Waiting Throughput
Model Length
Time (vehicles/hour)
(second) (vehicles)
Fixed -
Time
Signals 80 25 800
Fuzzy
Logic
Only 60 18 900
Genetic
Algorithm
Only 50 15 950
Genetic
Algorithm
- Fuzzy
Hybrid 40 12 1000
Table II. Comparison of Results (performance)

VI. REFERENCES
[1] N.C. Maduka, Ajibade and Bello, Modelling and Optimization of Smart Traffic Light Control System, 2022
IEEE Nigeria International Conference on Disruptive Technologies for Sustainable Development, (2022).
[2] Zhiyi Li, Mohammad Shahidehpur, Shay Bahramirad , Amin Khodaei , Optimizing traffic signal
settings in smart cities, IEEE Transactions on Smart Grid (2016).
[3] Anastasious Kouvelas , Markos , Elias, A hybrid strategy for real-time traffic signal control of urban road networks ,
IEEE Transactions on Intelligent Transportation Systems (2011).
[4] Mehrad Tajalli, Ali Hajbabaie, Traffic signal timing and trajectory optimization in a mixed autonomy traffic stream
, IEEE Transactions on Intelligent Transportation Systems (2021).
[5] Leena Singh, Sudhanshu Tripathi, Himakshi Arora, Time optimization for traffic signal control using genetic
algorithm, International Journal of Recent Trends in Engineering (2009 ).
[6] Dewasme, Urban Traffic Signals Timing at Four- Phase Signalized Intersection Based on Optimized Two-Stage
Fuzzy Control Scheme. Math. Problem Eng. 2021.
[7] Pitu Mirchandani , Larry Head, A real -time traffic signal control architecture, algorithms and analysis, Transportation
Research ; Emerging Technologies (2001).
[8] Saman Lawe, Ruili Wang , Optimization of traffic signals using deep learning neural networks , AI Joint Conference,
Hobert , TAS , Australia , (2016).
[9] Syed Shah Sultan Mohiuddin Qadri , State of art review of traffic signal control signals with challenges and
opportunities (2020).
[10] Yongji Jiang, Wei Jie Feng, Lei Wang, Xiangjie Kong, A rolling optimization algorithm for real time traffic control
with delay minimization.

[11] Stephen Chiu , Proceedings of the Intelligent Vehicles Symposium , Adaptive traffic signal system using fuzzy
logic, IEEE explore (2020)
[12] Dian Hartanti, Rosida Nur Aziza , Puji Catur Siswiprarptini, Optimization of smart traffic signals to prevent
traffic congestion using fuzzy logics , Telkomnika (2019).

[13] Sahar Araghi , Abbas Khosravi, Doug Creighton, Optimal design of traffic signal control using neural
networks and fuzzy logic systems, International Joint Conference (2014 ).
[14] Haifeng Lin, Yehong Han , Weiwei Cai , Bo Jin , Traffic signal optimization on the basis of fuzzy logic and
differential evolution algorithm , IEEE Transactions on Intelligent Transportation Systems (2022).
[15] Muzamil Etejani Mohammed Ali , Akif Durdu , Alper Yilmaz , An adaptive method for traffic signals system
on the basis of fuzzy logic and modified wester formula using SUMO traffic simulator , IEEE Access (2021).