Adaptive, Low-Cost Smart Traffic Management for Sana’a: A Context-Aware Approach to Urban Congestion in Low-Resource, Conflict-Affected Settings

Abstract

Urban congestion remains a persistent challenge in rapidly growing cities of developing countries, particularly in regions burdened by prolonged conflict and infrastructural neglect. This study investigates an adaptive, low-cost traffic signal control approach tailored for Sana’a, Yemen a city facing chronic traffic congestion, limited data infrastructure, and constrained public resources. Using the open-source traffic simulator SUMO (Simulation of Urban Mobility), we modeled a key intersection in Sana’a to evaluate the performance of two controllers: a traditional fixed-time signal and a context-aware adaptive controller that dynamically adjusts signal phases based on real-time vehicle density. The simulation integrates machine learning for congestion classification and real-time signal adaptation, producing performance metrics across vehicle throughput, queue length, and signal efficiency. Results reveal that the adaptive system significantly reduces congestion compared to the fixed-time controller, demonstrating better responsiveness to fluctuating traffic loads. Importantly, this solution requires no expensive hardware or centralized infrastructure—making it ideal for conflict-affected, low-resource urban settings. The findings underscore the potential of simulation-driven planning and low-cost AI methods in supporting smarter urban mobility, even in fragile cities. This work offers a scalable blueprint for traffic management in similar contexts across the Global South.

Introduction

Urban congestion is worsening in resource-constrained and conflict-affected cities like Sana’a, Yemen, where outdated traffic systems, poor infrastructure, and limited surveillance make traditional solutions infeasible. Intersections such as Al-Zubairi Street and Al-Hasaba face severe traffic delays, fuel waste, and pollution due to unregulated and rigid traffic signals.

2. Proposed Solution

This research introduces a low-cost, adaptive smart traffic light control system customized for Sana’a. Using the SUMO (Simulation of Urban Mobility) platform and machine learning, it simulates real-time signal control based on vehicle density and congestion prediction.

• Key Features:

  • Uses Random Forest classifiers for adaptive signal control.

  • Operates on basic inputs (e.g., low-cost traffic cameras or loop detectors).

  • Avoids expensive centralized infrastructure.

  • Provides interpretable and robust performance in low-data, high-noise environments.

3. Literature Review Insights

• Fixed-time signals are ineffective for dynamic traffic flows and worsen congestion in cities like Sana’a.

• Adaptive Traffic Signal Control (ATSC), enhanced by AI and simulation tools (like SUMO), significantly improves traffic throughput in high-tech cities—but less is known about low-resource applications.

• Random Forest models are preferred in sparse-data conditions due to their simplicity and resilience.

• Prior low-cost deployments in cities like Lagos and tier-2 Indian cities show feasibility on edge devices like Raspberry Pi.

• In conflict zones, systems must account for irregularities like sudden road closures, erratic driving, and degraded infrastructure.

4. Methodology

• Study Area: A high-traffic intersection in Sana’a with no automated sensing.

• SUMO Modeling:

  • Created using OpenStreetMap data and manual configurations.

  • Two traffic light programs were simulated: Fixed-time and Adaptive.

• Simulation Setup:

  • 10-minute sessions with 10-second data intervals.

  • Metrics: vehicle count, signal durations, predicted congestion.

  • Data stored in CSV logs and used for visual traffic analysis.

5. System Architecture

The system architecture simulates traffic flows, classifies congestion levels using trained ML models, and adjusts signal timings accordingly—all within the SUMO platform. The model loops continuously to adapt signal control in near-real time.

Conclusion

This study presents an adaptive and cost-effective approach to traffic signal management for Sana’a, a city facing chronic congestion in a low-resource, conflict-affected environment. By leveraging SUMO simulations, a Random Forest-based congestion prediction model, and fixed versus adaptive signal timing comparisons, the research has demonstrated the feasibility and advantages of intelligent traffic control in such constrained settings. The fixed-time traffic controller, commonly used in many developing countries, was compared with an adaptive controller integrated with congestion prediction logic. SUMO simulations illustrated that the adaptive system achieved better performance in managing vehicle accumulation and green light duration allocation. The simulation logs and plotted comparisons confirm a clear reduction in average vehicle presence and improved responsiveness of signal phases. Key findings include: • The adaptive model significantly reduced vehicle queue lengths at peak intervals. • Using local simulation data to retrain the ML model allowed for better context-specific predictions of congestion. • Fixed-timing controllers failed to respond to real-time traffic variation, leading to inefficient signal phases and longer delays. This work also illustrates the power of open-source tools like SUMO, joblib, pandas, and matplotlib in building full-stack traffic management systems without requiring commercial platforms or advanced infrastructure. The success of this pilot highlights how modest computational models can be deployed to enhance urban mobility even under conflict, lack of monitoring systems, and unreliable power.

References

[1] D. R. Aleko and S. Djahel, \"An Efficient Adaptive Traffic Light Control System for Urban Road Traffic Congestion Reduction in Smart Cities,\" Information, vol. 11, no. 2, pp. 1–21, 2020. [2] K. S. Mehta, K. N. Raj, and K. N. Brahmbhatt, \"Machine learning solutions for adaptive traffic signal control: A review of image?based approaches,\" World Journal of Advanced Engineering Technology and Sciences, vol. 13, no. 1, pp. 476–481, 2024. [3] X. Jia, M. Guo, Y. Lyu, J. Qu, D. Li, and F. Guo, \"Adaptive Traffic Signal Control Based on Graph Neural Networks and Dynamic Entropy-Constrained Soft Actor–Critic,\" Electronics, vol. 13, no. 23, pp. 1–20, 2024. [4] Anonymous, \"Adaptive Traffic Signal Control Algorithm Using Q learning with Loop Detector and CAV Data,\" Sustainability, vol. 14, no. 10, Art. no. 5751, pp. 1–18, 2021. [5] S. F. Smith, G. J. Barlow, and X. F. Xie, \"Smart and scalable urban signal networks,\" Proc. ACM SIGKDD Conf. Knowl. Discovery Data Mining, pp. 1–10, 2016. [6] D. R. Aleko and S. Djahel, An Efficient Adaptive Traffic Light Control System for Urban Road Traffic Congestion Reduction in Smart Cities, Information, vol. 11, no. 2, pp. 1–21, 2020. [7] K. Ye, A. Sharma, and M. Hasan, Multi-Agent Reinforcement Learning for Traffic Signal Control, ser. Lecture Notes in Computer Science. Berlin, Germany: Springer, 2023, vol. 13845. [8] H. Susilawati, S. L. P. Rahman, and A. Rukmana, Smart Traffic Light Using Raspberry Pi and Digital Image Processing, in Proc. 2023 9th Int. Conf. Wireless Telecommun. (ICWT), Kuala Lumpur, Malaysia, 2023, pp. 34–39. [9] D. Krajzewicz, G. Hertkorn, C. Rössel, and P. Wagner, Traffic Simulation with SUMO – Simulation of Urban MObility, in Fundamentals of Traffic Simulation, ser. International Series in Operations Research & Management Science. New York, NY, USA: Springer, 2010, vol. 145, pp. 1–20. [10] M. Wang, L. Zhang, and F. Liu, Traffic Data Imputation with MLP and Random Forest for Sparse Datasets, J. Adv. Transport., vol. 2022, pp. 1–13, 2022. [11] R. Singh, A. Mehta, and K. Verma, Edge-Based Traffic Signal Optimization Using Lightweight Random Forest Models, in Proc. 2023 Int. Conf. Embedded Intell. Syst. (EIS), Bhopal, India, 2023, pp. 90–95. [12] A. Okafor, L. Ayodele, and T. Ogunleye, Low-Cost Smart Traffic Light System Using Edge Devices: A Case Study in Lagos, Nigeria, Sustainability, vol. 13, no. 22, Art. no. 12891, 2021. [13] M. Elhadi and R. Alkhatib, Offline Retrainable AI Models for Traffic Signal Control in Conflict-Affected Zones, in Proc. 2022 Int. Conf. Emerging Smart Cities, Istanbul, Turkey, 2022, pp. 112–118. [14] Z. Mitkov, Civil War and Urban Infrastructure: Comparative Analysis of Traffic Management in Libya, Syria, and Yemen, M.S. thesis, Central European Univ., Budapest, Hungary, 2015.

Copyright

Copyright © 2025 Al-Madhaji Muhannad, B Kaveendran. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.