A systematic review of evolutionary and swarm intelligence approaches for traffic signal control optimization

Bhattacharyya, Rishika; Gupta, Sumit Kumar; Marisha; Kumar, Awadhesh; Mishra, Deepti; Gupta, Manjari · 2026 · OpenAlex-citations

DOI: 10.1007/s10462-026-11497-7

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This systematic review addresses the optimization of Traffic Signal Control (TSC) to mitigate urban congestion, travel delays, and emissions, particularly in developing nations with heterogeneous traffic streams. Motivated by the limitations of conventional static control systems in dynamic environments, the study evaluates the efficacy of Evolutionary Algorithms (EAs) and Swarm Intelligence (SI) methods. The authors aim to compare algorithmic performance, application contexts, and parameter tuning strategies to provide practical recommendations for researchers and policymakers. The study adheres to PRISMA guidelines, analyzing 50 peer-reviewed journal articles published between 2015 and 2025. The search strategy utilized databases including Web of Science, IEEE Xplore, and ScienceDirect, employing keywords related to traffic signal optimization, evolutionary algorithms, and swarm intelligence. Inclusion criteria restricted the review to original empirical research in English, excluding conference papers and book chapters to ensure methodological rigor. Studies were quality-assessed on a scale of 1 to 5, with those scoring below 3 excluded. Data extraction focused on algorithm types, datasets, performance metrics (e.g., average delay, emissions), and key outcomes, normalizing quantitative results for cross-study comparability. The findings indicate that hybrid methods, such as combining Genetic Algorithms (GA) with Particle Swarm Optimization (PSO), outperform single-algorithm approaches, achieving up to a 28.9% reduction in average vehicle delay. PSO demonstrates higher resilience for real-time applications, whereas GA offers robustness for offline, multi-objective planning. Specific parameter tuning significantly impacts performance, with optimal GA mutation rates identified between 0.01 and 0.1, and PSO inertia coefficients around 0.7. Variants like LDW-PSO reduced premature convergence by 15%, and ACO integrated with IoT sensors improved route optimization accuracy by 30%. Hybrid models, such as ACO-PSO, achieved 30% higher route optimization accuracy, though often requiring costly infrastructure like 5G/RFID. The review concludes that hybridization is critical for handling complex, dynamic traffic scenarios effectively. While Reinforcement Learning (RL) shows competitive adaptability, hybrid EA-SI frameworks currently offer superior convergence stability and interpretability. The study highlights a trade-off between performance gains and implementation costs, noting that advanced hybrid systems often require significant infrastructure investment. These insights provide a synthesized evidence base for deploying adaptive traffic management systems, emphasizing the need for scalable, equity-aware optimization strategies that balance computational efficiency with real-world applicability.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• regulatory evaluation