Weighted Random Early Detection in Modern Active Queue Management: A Structured Literature Review (2002–2025)

Abstract

Active Queue Management (AQM) mechanisms are critical for mitigating congestion and maintaining quality of service in packet-switched networks. Weighted Random Early Detection (WRED) remains widely deployed due to its support for traffic differentiation within DiffServ architectures. However, the broader AQM landscape has evolved to include adaptive, rate-based, delay-based, and learning-driven algorithms designed to address limitations of early queue-length-based approaches. This study presents a structured literature review examining WRED’s position within modern AQM frameworks from 2002 to 2025. A systematic search across major academic databases yielded 168 records, from which 34 studies were selected for detailed synthesis following eligibility screening. Comparative analysis indicates that WRED effectively preserves priority traffic under moderate congestion but remains sensitive to parameter configuration and may degrade fairness under sustained load. In contrast, adaptive and delay-based AQMs improve stability and latency control but lack intrinsic service differentiation. The findings highlight WRED’s continued relevance in QoS-aware environments while emphasizing the need for adaptive and fairness-aware enhancements to improve robustness in heterogeneous network conditions.

DOI: https://doi.org/10.5281/zenodo.19090741