Artificial Intelligence Driven Congestion Aware Routing Framework for Next Generation Communication Networks

With the exponential rise in connected devices and data-heavy applications, traditional routing protocols struggle to efficiently manage congestion in next-generation communication networks. In this paper I introduce a congestion-aware routing framework (AI-CARF) that was developed based on Artificial Intelligence (AI) to proactively control the impact of congestion through a multi-layered intelligence framework. The framework includes Graph Signal Processing to encode topology, Topological Data Analysis to identify anomalies using persistence, and Neuroevolution to develop context-based routing capabilities. Moreover, Fuzzy Reinforcement Learning model is used to interpret quality-of-service measures, whereas Multi-Agent Deep Q-Learning helps to make decentralized and real-time decisions. Knowledge distillation is done in a simulated digital twin environment, which enables lightweight deployment of the model in edge devices. The model was experimented with various synthetic and real-world traffic topologies and demonstrated a congestion-aware routing quality of 96.42 which is better than the conventional AODV and DSR protocols in terms of packet delivery ratio, delay, and jitter. Findings verify the strength of AI-CARF when a load is varied, a node failed and topology changed dynamically. The framework is a breakthrough towards autonomous, scalable and adaptive routing in ultra-dense communication networks like 6G, IoT mesh networks, VANETs and satellite communications, which will enable intelligent infrastructure optimization in network generations to come.