LAMP5G: A Link-Aware Multipath QUIC Scheduler for Heterogeneous and Disruption-Prone 5G Cellular Networks

Multipath transport has emerged as a promising solution to address key challenges in 5G cellular networks, including improved performance with limited RAN resources, resilience against link failures, and operator heterogeneity. While emerging 5G architectures such as NR dual connectivity (NR-DC) and multi-operator deployments potentially enable simultaneous access to multiple paths, efficiently utilizing these paths remains challenging due to dynamic link conditions, frequent handovers, and environmental impairments.

In this paper, we present LAMP5G, a link-aware scheduler built on Multipath QUIC (MPQUIC) for 5G cellular networks. LAMP5G employs a dynamic path-ranking mechanism designed for heterogeneous path environments. The ranking algorithm leverages both end-to-end transport-layer metrics—such as throughput, round-trip time (RTT), loss rate, and congestion window size—and local link-layer statistics to estimate available path bandwidth. To mitigate disruptions caused by handovers, LAMP5G implements a make-before-break strategy through explicit path management, enabling seamless traffic migration between RANs.

We evaluate LAMP5G on a 5G testbed. Experimental results show that LAMP5G reduces completion time by 57% compared to the dual-QUIC baseline under heterogeneous path conditions. Furthermore, during mobility scenarios, LAMP5G attains approximately 60% higher average throughput compared to traditional handover.