A QoS-driven Fuzzy Logic-based Adaptive MPQUIC Scheduler for Heterogeneous Networks

Multipath communication has become a fundamental technique for improving network performance and reliability. However, in heterogeneous and dynamic network environments, conventional scheduling mechanisms often suffer from issues such as head-of-line (HOL) blocking and receiver buffer overflow, leading to suboptimal transmission efficiency. To address the limitations, we propose Stream-Aware Optimal Path First (SA-OPF), a novel adaptive scheduler for Multipath QUIC (MPQUIC) that dynamically selects the optimal path for each packet transmission. Firstly, SA-OPF allocates different transmission resources to different streams based on stream weights to meet quality of service (QoS) requirements, known as stream-aware scheduling. Then, SA-OPF leverages a fuzzy logic system to evaluate the performance of each path in real-time, ensuring that packets are forwarded along the optimal path, known as adaptive scheduling policy. By incorporating stream-aware and adaptive optimal path scheduling, SA-OPF improves QoS for high-priority streams while providing a minimum bandwidth guarantee for low-priority streams. We implement SA-OPF in MPQUIC and evaluate its performance in various heterogeneous network scenarios in both emulated and real-world network environments. Experimental results demonstrate that SA-OPF significantly improves data transmission efficiency and reduces transmission latency compared to existing MPQUIC schedulers.