A hybrid bio-inspired and machine learning framework for adaptive congestion control and multi-objective QoS optimization in heterogeneous network

Heterogeneous communication networks, that is, the integration of wireless access networks (IoT, Wireless Sensor Networks, Mobile Ad Hoc Networks, Vehicular Ad Hoc Networks, and 5G/6G terrestrial networks), software-defined and cloud-edge infrastructures, and satellite-assisted non-terrestrial networks (NTN), present a serious challenge to adaptive congestion control because of the diversity of technologies, topologies, mobility patterns, and QoS requirements. Although there are various solutions to this problem, such as bio-inspired optimisation, machine-learning prediction, and reinforcement-learning control, none of them addresses the whole problem by considering a combination of global path exploration, proactive congestion forecasting, and adaptive routing control. We propose a 4-layer hybrid framework, CAIRES-Arterial, which combines a genetic algorithm (GA) for Global Path Exploration, Ant Colony Optimisation (ACO) for Pheromone-Guided Path Refining, CatBoost Gradient Boosting for Proactive Congestion State Prediction, and Q-learning for Decentralised Node-level Adaptive Forwarding. The confidence-weighted Hybrid Decision Engine combines GA fitness score, ACO pheromone strength, CatBoost prediction confidence, and Q-learning state-action values to make composite routing decisions in response to real-time network conditions. Extensive simulations with synthetic and real traffic traces (MAWI/CAIDA), using NS-3, GNS3 and Mininet, random, grid and scale-free topologies, show that CAIRES-Arterial outperforms both GA-ACO, CatBoost-Q-learning, DQN and OSPF baselines by 8–15% in terms of throughput, > 92% packet delivery ratio, up to 18% energy saving and up to 35% faster congestion recovery (statistically significant, p < 0.01), while being containerised and deployed in an 512-node topology.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-17
DOI
https://doi.org/10.1038/s41598-026-77348-1
Primary Topic
Network Traffic and Congestion Control
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A hybrid bio-inspired and machine learning framework for adaptive congestion control and multi-objective QoS optimization in heterogeneous network

T. Sampath Kumar, Sripelli Jagadish
1 links
Scientific Reports
Network Traffic and Congestion Control
article

A hybrid bio-inspired and machine learning framework for adaptive congestion control and multi-objective QoS optimization in heterogeneous network

T. Sampath Kumar, Sripelli Jagadish
article en

Abstract

Heterogeneous communication networks, that is, the integration of wireless access networks (IoT, Wireless Sensor Networks, Mobile Ad Hoc Networks, Vehicular Ad Hoc Networks, and 5G/6G terrestrial networks), software-defined and cloud-edge infrastructures, and satellite-assisted non-terrestrial networks (NTN), present a serious challenge to adaptive congestion control because of the diversity of technologies, topologies, mobility patterns, and QoS requirements. Although there are various solutions to this problem, such as bio-inspired optimisation, machine-learning prediction, and reinforcement-learning control, none of them addresses the whole problem by considering a combination of global path exploration, proactive congestion forecasting, and adaptive routing control. We propose a 4-layer hybrid framework, CAIRES-Arterial, which combines a genetic algorithm (GA) for Global Path Exploration, Ant Colony Optimisation (ACO) for Pheromone-Guided Path Refining, CatBoost Gradient Boosting for Proactive Congestion State Prediction, and Q-learning for Decentralised Node-level Adaptive Forwarding. The confidence-weighted Hybrid Decision Engine combines GA fitness score, ACO pheromone strength, CatBoost prediction confidence, and Q-learning state-action values to make composite routing decisions in response to real-time network conditions. Extensive simulations with synthetic and real traffic traces (MAWI/CAIDA), using NS-3, GNS3 and Mininet, random, grid and scale-free topologies, show that CAIRES-Arterial outperforms both GA-ACO, CatBoost-Q-learning, DQN and OSPF baselines by 8–15% in terms of throughput, > 92% packet delivery ratio, up to 18% energy saving and up to 35% faster congestion recovery (statistically significant, p < 0.01), while being containerised and deployed in an 512-node topology.

Scientific Reports
SR University (IN)
Industry, innovation and infrastructure
Network Traffic and Congestion Control
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon